phase 10

core/api/detect_replay.py

@@ -0,0 +1,121 @@
+"""
+API endpoints for checkpoint inspection, replay, and retry.
+
+GET  /detect/checkpoints/{job_id}       — list available checkpoints
+POST /detect/replay                     — replay from a stage with config overrides
+POST /detect/retry                      — queue async retry with different provider
+"""
+
+from __future__ import annotations
+
+import logging
+
+from fastapi import APIRouter, HTTPException
+from pydantic import BaseModel
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter(prefix="/detect", tags=["detect"])
+
+
+# --- Request/Response models ---
+
+class CheckpointInfo(BaseModel):
+    stage: str
+
+
+class ReplayRequest(BaseModel):
+    job_id: str
+    start_stage: str
+    config_overrides: dict | None = None
+
+
+class ReplayResponse(BaseModel):
+    status: str
+    job_id: str
+    start_stage: str
+    detections: int = 0
+    brands_found: int = 0
+
+
+class RetryRequest(BaseModel):
+    job_id: str
+    config_overrides: dict | None = None
+    start_stage: str = "escalate_vlm"
+    schedule_seconds: float | None = None  # delay before execution (off-peak)
+
+
+class RetryResponse(BaseModel):
+    status: str
+    task_id: str
+    job_id: str
+
+
+# --- Endpoints ---
+
+@router.get("/checkpoints/{job_id}")
+def list_checkpoints(job_id: str) -> list[CheckpointInfo]:
+    """List available checkpoint stages for a job."""
+    from detect.checkpoint import list_checkpoints as _list
+
+    try:
+        stages = _list(job_id)
+    except Exception as e:
+        raise HTTPException(status_code=404, detail=f"No checkpoints for job {job_id}: {e}")
+
+    result = [CheckpointInfo(stage=s) for s in stages]
+    return result
+
+
+@router.post("/replay", response_model=ReplayResponse)
+def replay(req: ReplayRequest):
+    """Replay pipeline from a specific stage with optional config overrides."""
+    from detect.checkpoint import replay_from
+
+    try:
+        result = replay_from(
+            job_id=req.job_id,
+            start_stage=req.start_stage,
+            config_overrides=req.config_overrides,
+        )
+    except ValueError as e:
+        raise HTTPException(status_code=400, detail=str(e))
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Replay failed: {e}")
+
+    detections = result.get("detections", [])
+    report = result.get("report")
+    brands_found = len(report.brands) if report else 0
+
+    response = ReplayResponse(
+        status="completed",
+        job_id=req.job_id,
+        start_stage=req.start_stage,
+        detections=len(detections),
+        brands_found=brands_found,
+    )
+    return response
+
+
+@router.post("/retry", response_model=RetryResponse)
+def retry(req: RetryRequest):
+    """Queue an async retry of unresolved candidates with different config."""
+    from detect.checkpoint.tasks import retry_candidates
+
+    kwargs = {
+        "job_id": req.job_id,
+        "config_overrides": req.config_overrides,
+        "start_stage": req.start_stage,
+    }
+
+    if req.schedule_seconds:
+        task = retry_candidates.apply_async(kwargs=kwargs, countdown=req.schedule_seconds)
+    else:
+        task = retry_candidates.delay(**kwargs)
+
+    response = RetryResponse(
+        status="queued",
+        task_id=task.id,
+        job_id=req.job_id,
+    )
+    return response

core/api/main.py

@@ -25,6 +25,7 @@ from strawberry.fastapi import GraphQLRouter
 
 from core.api.chunker_sse import router as chunker_router
 from core.api.detect_sse import router as detect_router
+from core.api.detect_replay import router as detect_replay_router
 from core.api.graphql import schema as graphql_schema
 
 CALLBACK_API_KEY = os.environ.get("CALLBACK_API_KEY", "")
@@ -56,6 +57,9 @@ app.include_router(chunker_router)
 # Detection SSE
 app.include_router(detect_router)
 
+# Detection replay/retry
+app.include_router(detect_replay_router)
+
 
 @app.get("/health")
 def health():

core/db/detect.py

@@ -0,0 +1,175 @@
+"""Database operations for DetectJob and StageCheckpoint."""
+
+from typing import Optional
+from uuid import UUID
+
+
+# ---------------------------------------------------------------------------
+# DetectJob
+# ---------------------------------------------------------------------------
+
+def create_detect_job(**fields):
+    from admin.mpr.media_assets.models import DetectJob
+    return DetectJob.objects.create(**fields)
+
+
+def get_detect_job(id: UUID):
+    from admin.mpr.media_assets.models import DetectJob
+    return DetectJob.objects.get(id=id)
+
+
+def update_detect_job(job_id: UUID, **fields):
+    from admin.mpr.media_assets.models import DetectJob
+    DetectJob.objects.filter(id=job_id).update(**fields)
+
+
+def list_detect_jobs(
+    parent_job_id: Optional[UUID] = None,
+    status: Optional[str] = None,
+):
+    from admin.mpr.media_assets.models import DetectJob
+
+    qs = DetectJob.objects.all()
+    if parent_job_id:
+        qs = qs.filter(parent_job_id=parent_job_id)
+    if status:
+        qs = qs.filter(status=status)
+    return list(qs)
+
+
+# ---------------------------------------------------------------------------
+# StageCheckpoint
+# ---------------------------------------------------------------------------
+
+def save_stage_checkpoint(**fields):
+    from admin.mpr.media_assets.models import StageCheckpoint
+    return StageCheckpoint.objects.create(**fields)
+
+
+def get_stage_checkpoint(job_id: UUID, stage: str):
+    from admin.mpr.media_assets.models import StageCheckpoint
+    return StageCheckpoint.objects.get(job_id=job_id, stage=stage)
+
+
+def list_stage_checkpoints(job_id: UUID) -> list[str]:
+    from admin.mpr.media_assets.models import StageCheckpoint
+
+    stages = (
+        StageCheckpoint.objects
+        .filter(job_id=job_id)
+        .order_by("stage_index")
+        .values_list("stage", flat=True)
+    )
+    return list(stages)
+
+
+def delete_stage_checkpoints(job_id: UUID):
+    from admin.mpr.media_assets.models import StageCheckpoint
+    StageCheckpoint.objects.filter(job_id=job_id).delete()
+
+
+# ---------------------------------------------------------------------------
+# KnownBrand
+# ---------------------------------------------------------------------------
+
+def get_or_create_brand(canonical_name: str, aliases: list[str] | None = None,
+                        source: str = "ocr") -> tuple:
+    """Get existing brand or create new one. Returns (brand, created)."""
+    from admin.mpr.media_assets.models import KnownBrand
+    import uuid
+
+    normalized = canonical_name.strip()
+    brand = KnownBrand.objects.filter(canonical_name__iexact=normalized).first()
+    if brand:
+        return brand, False
+
+    # Check aliases of existing brands
+    for existing in KnownBrand.objects.all():
+        existing_aliases = [a.lower() for a in (existing.aliases or [])]
+        if normalized.lower() in existing_aliases:
+            return existing, False
+
+    brand = KnownBrand.objects.create(
+        id=uuid.uuid4(),
+        canonical_name=normalized,
+        aliases=aliases or [],
+        first_source=source,
+    )
+    return brand, True
+
+
+def find_brand_by_text(text: str) -> Optional[object]:
+    """Find a known brand by canonical name or alias (case-insensitive)."""
+    from admin.mpr.media_assets.models import KnownBrand
+
+    normalized = text.strip().lower()
+
+    # Exact canonical match
+    brand = KnownBrand.objects.filter(canonical_name__iexact=normalized).first()
+    if brand:
+        return brand
+
+    # Search aliases (jsonb contains)
+    for brand in KnownBrand.objects.all():
+        brand_aliases = [a.lower() for a in (brand.aliases or [])]
+        if normalized in brand_aliases:
+            return brand
+
+    return None
+
+
+def list_all_brands() -> list:
+    from admin.mpr.media_assets.models import KnownBrand
+    return list(KnownBrand.objects.all().order_by("canonical_name"))
+
+
+def update_brand(brand_id: UUID, **fields):
+    from admin.mpr.media_assets.models import KnownBrand
+    KnownBrand.objects.filter(id=brand_id).update(**fields)
+
+
+# ---------------------------------------------------------------------------
+# SourceBrandSighting
+# ---------------------------------------------------------------------------
+
+def get_source_sightings(source_asset_id: UUID) -> list:
+    """Get all brand sightings for a specific source video."""
+    from admin.mpr.media_assets.models import SourceBrandSighting
+    return list(
+        SourceBrandSighting.objects
+        .filter(source_asset_id=source_asset_id)
+        .order_by("-occurrences")
+    )
+
+
+def record_sighting(source_asset_id: UUID, brand_id: UUID, brand_name: str,
+                    timestamp: float, confidence: float, source: str = "ocr"):
+    """Record or update a brand sighting for a source."""
+    from admin.mpr.media_assets.models import SourceBrandSighting
+    import uuid
+
+    sighting = SourceBrandSighting.objects.filter(
+        source_asset_id=source_asset_id,
+        brand_id=brand_id,
+    ).first()
+
+    if sighting:
+        sighting.occurrences += 1
+        sighting.last_seen_timestamp = timestamp
+        total_conf = sighting.avg_confidence * (sighting.occurrences - 1) + confidence
+        sighting.avg_confidence = total_conf / sighting.occurrences
+        sighting.save()
+        return sighting
+
+    sighting = SourceBrandSighting.objects.create(
+        id=uuid.uuid4(),
+        source_asset_id=source_asset_id,
+        brand_id=brand_id,
+        brand_name=brand_name,
+        first_seen_timestamp=timestamp,
+        last_seen_timestamp=timestamp,
+        occurrences=1,
+        detection_source=source,
+        avg_confidence=confidence,
+    )
+    return sighting

core/schema/models/__init__.py

@@ -26,13 +26,18 @@ from .grpc import (
     WorkerStatus,
 )
 from .jobs import ChunkJob, ChunkJobStatus, JobStatus, TranscodeJob
+from .detect_jobs import (
+    DetectJob, DetectJobStatus, RunType, StageCheckpoint,
+    BrandSource, KnownBrand, SourceBrandSighting,
+)
 from .media import AssetStatus, MediaAsset
 from .presets import BUILTIN_PRESETS, TranscodePreset
 from .detect import DETECT_VIEWS  # noqa: F401 — discovered by modelgen generic loader
 from .views import ChunkEvent, ChunkOutputFile, PipelineStats, WorkerEvent
 
 # Core domain models - generates Django, Pydantic, TypeScript
-DATACLASSES = [MediaAsset, TranscodePreset, TranscodeJob, ChunkJob]
+DATACLASSES = [MediaAsset, TranscodePreset, TranscodeJob, ChunkJob,
+               DetectJob, StageCheckpoint, KnownBrand, SourceBrandSighting]
 
 # API request/response models - generates TypeScript only (no Django)
 # WorkerStatus from grpc.py is reused here
@@ -46,7 +51,7 @@ API_MODELS = [
 ]
 
 # Status enums - included in generated code
-ENUMS = [AssetStatus, JobStatus, ChunkJobStatus]
+ENUMS = [AssetStatus, JobStatus, ChunkJobStatus, DetectJobStatus, RunType, BrandSource]
 
 # View/event models - generates TypeScript for UI consumption
 VIEWS = [ChunkEvent, WorkerEvent, PipelineStats, ChunkOutputFile]

core/schema/models/detect.py

@@ -149,6 +149,64 @@ class JobComplete:
     report: Optional[DetectionReportSummary] = None
 
 
+@dataclass
+class RunContext:
+    """Run context injected into all SSE events for grouping."""
+
+    run_id: str
+    parent_job_id: str
+    run_type: str = "initial"  # initial | replay | retry
+
+
+# --- Checkpoint API types ---
+
+
+@dataclass
+class CheckpointInfo:
+    """Available checkpoint for a stage."""
+
+    stage: str
+
+
+@dataclass
+class ReplayRequest:
+    """Request to replay pipeline from a specific stage."""
+
+    job_id: str
+    start_stage: str
+    config_overrides: Optional[dict] = None
+
+
+@dataclass
+class ReplayResponse:
+    """Result of a replay invocation."""
+
+    status: str
+    job_id: str
+    start_stage: str
+    detections: int = 0
+    brands_found: int = 0
+
+
+@dataclass
+class RetryRequest:
+    """Request to queue async retry with different config."""
+
+    job_id: str
+    config_overrides: Optional[dict] = None
+    start_stage: str = "escalate_vlm"
+    schedule_seconds: Optional[float] = None
+
+
+@dataclass
+class RetryResponse:
+    """Result of queueing a retry task."""
+
+    status: str
+    task_id: str
+    job_id: str
+
+
 # --- Export lists for modelgen ---
 
 DETECT_VIEWS = [
@@ -163,4 +221,10 @@ DETECT_VIEWS = [
     LogEvent,
     DetectionReportSummary,
     JobComplete,
+    RunContext,
+    CheckpointInfo,
+    ReplayRequest,
+    ReplayResponse,
+    RetryRequest,
+    RetryResponse,
 ]

core/schema/models/detect_jobs.py

@@ -0,0 +1,162 @@
+"""
+Detection Job and Checkpoint Schema Definitions
+
+Source of truth for detection pipeline job tracking and stage checkpoints.
+Follows the TranscodeJob/ChunkJob pattern.
+"""
+
+from dataclasses import dataclass, field
+from datetime import datetime
+from enum import Enum
+from typing import Any, Dict, List, Optional
+from uuid import UUID
+
+
+class DetectJobStatus(str, Enum):
+    PENDING = "pending"
+    RUNNING = "running"
+    PAUSED = "paused"
+    COMPLETED = "completed"
+    FAILED = "failed"
+    CANCELLED = "cancelled"
+
+
+class RunType(str, Enum):
+    INITIAL = "initial"
+    REPLAY = "replay"
+    RETRY = "retry"
+
+
+@dataclass
+class DetectJob:
+    """
+    A detection pipeline job.
+
+    Each invocation of the pipeline (initial run, replay, retry) creates a DetectJob.
+    Jobs for the same source video are linked via parent_job_id.
+    """
+
+    id: UUID
+
+    # Input
+    source_asset_id: UUID
+    video_path: str
+    profile_name: str = "soccer_broadcast"
+
+    # Run lineage
+    parent_job_id: Optional[UUID] = None  # links all runs for the same source
+    run_type: RunType = RunType.INITIAL
+    replay_from_stage: Optional[str] = None  # null for initial runs
+    config_overrides: Dict[str, Any] = field(default_factory=dict)
+
+    # Status
+    status: DetectJobStatus = DetectJobStatus.PENDING
+    current_stage: Optional[str] = None
+    progress: float = 0.0
+    error_message: Optional[str] = None
+
+    # Results summary
+    total_detections: int = 0
+    brands_found: int = 0
+    cloud_llm_calls: int = 0
+    estimated_cost_usd: float = 0.0
+
+    # Worker tracking
+    celery_task_id: Optional[str] = None
+    priority: int = 0
+
+    # Timestamps
+    created_at: Optional[datetime] = None
+    started_at: Optional[datetime] = None
+    completed_at: Optional[datetime] = None
+
+
+@dataclass
+class StageCheckpoint:
+    """
+    A checkpoint saved after a pipeline stage completes.
+
+    Binary data (frame images, crops) goes to S3/MinIO.
+    Everything else (structured state) lives here in Postgres.
+    """
+
+    id: UUID
+    job_id: UUID
+    stage: str
+    stage_index: int  # position in NODES list (0-7)
+
+    # S3 reference for binary data only
+    frames_prefix: str = ""  # s3 prefix: checkpoints/{job_id}/frames/
+
+    # Frame metadata (non-image fields)
+    frames_manifest: Dict[int, str] = field(default_factory=dict)  # seq → s3 key
+    frames_meta: List[Dict[str, Any]] = field(default_factory=list)  # sequence, chunk_id, timestamp, hash
+    filtered_frame_sequences: List[int] = field(default_factory=list)
+
+    # Detection state (full structured data, not just summaries)
+    boxes_by_frame: Dict[str, List[Dict[str, Any]]] = field(default_factory=dict)
+    text_candidates: List[Dict[str, Any]] = field(default_factory=list)
+    unresolved_candidates: List[Dict[str, Any]] = field(default_factory=list)
+    detections: List[Dict[str, Any]] = field(default_factory=list)
+
+    # Pipeline state
+    stats: Dict[str, Any] = field(default_factory=dict)
+    config_snapshot: Dict[str, Any] = field(default_factory=dict)
+    config_overrides: Dict[str, Any] = field(default_factory=dict)
+
+    # Input refs (for replay)
+    video_path: str = ""
+    profile_name: str = ""
+
+    # Timestamps
+    created_at: Optional[datetime] = None
+
+
+class BrandSource(str, Enum):
+    """How a brand was first identified."""
+    OCR = "ocr"
+    VLM = "local_vlm"
+    CLOUD = "cloud_llm"
+    MANUAL = "manual"  # user-added via UI
+
+
+@dataclass
+class KnownBrand:
+    """
+    A brand discovered or registered in the system.
+
+    Global — not per-source. Accumulates across all pipeline runs.
+    Aliases enable fuzzy matching without re-escalating to VLM.
+    """
+
+    id: UUID
+    canonical_name: str              # normalized display name
+    aliases: List[str] = field(default_factory=list)  # known spellings/variants
+    first_source: BrandSource = BrandSource.OCR
+    total_occurrences: int = 0
+    confirmed: bool = False          # manually confirmed by user
+
+    created_at: Optional[datetime] = None
+    updated_at: Optional[datetime] = None
+
+
+@dataclass
+class SourceBrandSighting:
+    """
+    A brand seen in a specific source (video/asset).
+
+    Per-source session cache — avoids re-escalating the same brand
+    on subsequent frames or re-runs of the same source.
+    """
+
+    id: UUID
+    source_asset_id: UUID            # the video this sighting belongs to
+    brand_id: UUID                   # FK to KnownBrand
+    brand_name: str                  # denormalized for fast lookup
+    first_seen_timestamp: float = 0.0
+    last_seen_timestamp: float = 0.0
+    occurrences: int = 0
+    detection_source: BrandSource = BrandSource.OCR
+    avg_confidence: float = 0.0
+
+    created_at: Optional[datetime] = None

ctrl/.env.template

@@ -28,7 +28,10 @@ GRPC_PORT=50051
 GRPC_MAX_WORKERS=10
 
 # S3 Storage (MinIO locally, real S3 on AWS)
-S3_ENDPOINT_URL=http://minio:9000
+# In k8s/docker: http://minio:9000
+# On dev machine (port-forward): http://localhost:9000
+# On AWS: omit S3_ENDPOINT_URL entirely
+S3_ENDPOINT_URL=http://localhost:9000
 S3_BUCKET_IN=mpr-media-in
 S3_BUCKET_OUT=mpr-media-out
 AWS_REGION=us-east-1
@@ -44,7 +47,7 @@ CLOUD_LLM_PROVIDER=groq
 
 # Groq (default, free tier)
 GROQ_API_KEY=
-GROQ_MODEL=llama-3.2-90b-vision-preview
+GROQ_MODEL=meta-llama/llama-4-scout-17b-16e-instruct
 
 # Gemini
 #GEMINI_API_KEY=

ctrl/Tiltfile

@@ -35,7 +35,15 @@ docker_build(
 # --- Resources ---
 
 k8s_resource('redis')
-k8s_resource('fastapi', resource_deps=['redis'])
+k8s_resource('minio', port_forwards=['9000:9000', '9001:9001'])
+k8s_resource('fastapi', resource_deps=['redis', 'minio'])
 k8s_resource('detection-ui')
 k8s_resource('gateway', resource_deps=['fastapi', 'detection-ui'],
              port_forwards=['8080:8080'])
+
+# Group uncategorized resources (configmaps, namespace) under infra
+k8s_resource(
+    objects=['mpr:namespace', 'mpr-config:configmap', 'minio-config:configmap',
+             'envoy-gateway-config:configmap'],
+    new_name='infra',
+)

ctrl/k8s/base/fastapi.yaml

@@ -22,6 +22,8 @@ spec:
           envFrom:
             - configMapRef:
                 name: mpr-config
+            - configMapRef:
+                name: minio-config
           readinessProbe:
             httpGet:
               path: /health

ctrl/k8s/base/kustomization.yaml

@@ -7,6 +7,7 @@ resources:
   - namespace.yaml
   - configmap.yaml
   - redis.yaml
+  - minio.yaml
   - fastapi.yaml
   - detection-ui.yaml
   - gateway.yaml

ctrl/k8s/base/minio.yaml

@@ -0,0 +1,87 @@
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: minio-config
+  namespace: mpr
+data:
+  S3_ENDPOINT_URL: http://minio:9000
+  S3_BUCKET_IN: mpr-media-in
+  S3_BUCKET_OUT: mpr-media-out
+  AWS_ACCESS_KEY_ID: minioadmin
+  AWS_SECRET_ACCESS_KEY: minioadmin
+  AWS_REGION: us-east-1
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: minio
+  namespace: mpr
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: minio
+  template:
+    metadata:
+      labels:
+        app: minio
+    spec:
+      containers:
+        - name: minio
+          image: minio/minio:latest
+          args: ["server", "/data", "--console-address", ":9001"]
+          ports:
+            - containerPort: 9000
+              name: api
+            - containerPort: 9001
+              name: console
+          env:
+            - name: MINIO_ROOT_USER
+              value: minioadmin
+            - name: MINIO_ROOT_PASSWORD
+              value: minioadmin
+          readinessProbe:
+            httpGet:
+              path: /minio/health/ready
+              port: 9000
+            initialDelaySeconds: 5
+            periodSeconds: 10
+          lifecycle:
+            postStart:
+              exec:
+                command:
+                  - /bin/sh
+                  - -c
+                  - |
+                    sleep 3
+                    for bucket in mpr-media-in mpr-media-out; do
+                      mkdir -p /data/$bucket
+                    done
+          volumeMounts:
+            - name: data
+              mountPath: /data
+          resources:
+            requests:
+              memory: 128Mi
+              cpu: 100m
+            limits:
+              memory: 512Mi
+      volumes:
+        - name: data
+          emptyDir: {}
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: minio
+  namespace: mpr
+spec:
+  selector:
+    app: minio
+  ports:
+    - port: 9000
+      targetPort: 9000
+      name: api
+    - port: 9001
+      targetPort: 9001
+      name: console

detect/checkpoint/__init__.py

@@ -0,0 +1,14 @@
+"""
+Stage checkpoint, replay, and retry.
+
+    detect/checkpoint/
+        frames.py      — frame image S3 upload/download
+        serializer.py  — state ↔ JSON conversion
+        storage.py     — checkpoint save/load/list (Postgres + S3)
+        replay.py      — replay_from, OverrideProfile
+        tasks.py       — retry_candidates Celery task
+"""
+
+from .storage import save_checkpoint, load_checkpoint, list_checkpoints
+from .frames import save_frames, load_frames
+from .replay import replay_from, OverrideProfile

detect/checkpoint/frames.py

@@ -0,0 +1,80 @@
+"""Frame image storage — save/load to S3/MinIO as JPEGs."""
+
+from __future__ import annotations
+
+import logging
+import os
+import tempfile
+
+import numpy as np
+from PIL import Image
+
+from detect.models import Frame
+
+logger = logging.getLogger(__name__)
+
+BUCKET = os.environ.get("S3_BUCKET_OUT", "mpr-media-out")
+CHECKPOINT_PREFIX = "checkpoints"
+
+
+def save_frames(job_id: str, frames: list[Frame]) -> dict[int, str]:
+    """
+    Save frame images to S3 as JPEGs.
+
+    Returns manifest: {sequence: s3_key}
+    """
+    from core.storage.s3 import upload_file
+
+    manifest = {}
+
+    for frame in frames:
+        key = f"{CHECKPOINT_PREFIX}/{job_id}/frames/{frame.sequence}.jpg"
+
+        with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as tmp:
+            img = Image.fromarray(frame.image)
+            img.save(tmp, format="JPEG", quality=85)
+            tmp_path = tmp.name
+
+        try:
+            upload_file(tmp_path, BUCKET, key)
+        finally:
+            os.unlink(tmp_path)
+
+        manifest[frame.sequence] = key
+
+    logger.info("Saved %d frames to s3://%s/%s/%s/frames/",
+                len(frames), BUCKET, CHECKPOINT_PREFIX, job_id)
+    return manifest
+
+
+def load_frames(manifest: dict[int, str], frame_metadata: list[dict]) -> list[Frame]:
+    """
+    Load frame images from S3 and reconstitute Frame objects.
+
+    frame_metadata: list of dicts with sequence, chunk_id, timestamp, perceptual_hash.
+    """
+    from core.storage.s3 import download_to_temp
+
+    meta_map = {m["sequence"]: m for m in frame_metadata}
+    frames = []
+
+    for seq, key in manifest.items():
+        tmp_path = download_to_temp(BUCKET, key)
+        try:
+            img = Image.open(tmp_path).convert("RGB")
+            image_array = np.array(img)
+        finally:
+            os.unlink(tmp_path)
+
+        meta = meta_map.get(seq, {})
+        frame = Frame(
+            sequence=seq,
+            chunk_id=meta.get("chunk_id", 0),
+            timestamp=meta.get("timestamp", 0.0),
+            image=image_array,
+            perceptual_hash=meta.get("perceptual_hash", ""),
+        )
+        frames.append(frame)
+
+    frames.sort(key=lambda f: f.sequence)
+    return frames

detect/checkpoint/replay.py

@@ -0,0 +1,132 @@
+"""
+Pipeline replay — re-run from any stage with different config.
+
+Loads a checkpoint, applies config overrides, builds a subgraph
+starting from the target stage, and invokes it.
+"""
+
+from __future__ import annotations
+
+import logging
+
+import uuid
+
+from detect import emit
+from detect.checkpoint import load_checkpoint, list_checkpoints
+from detect.graph import NODES, build_graph
+
+logger = logging.getLogger(__name__)
+
+
+class OverrideProfile:
+    """
+    Wraps a ContentTypeProfile and patches config methods with overrides.
+
+    Override dict structure:
+        {
+            "frame_extraction": {"fps": 1.0},
+            "scene_filter": {"hamming_threshold": 12},
+            "detection": {"confidence_threshold": 0.5},
+            "ocr": {"languages": ["en", "es"], "min_confidence": 0.3},
+            "resolver": {"fuzzy_threshold": 60},
+        }
+    """
+
+    def __init__(self, base, overrides: dict):
+        self._base = base
+        self._overrides = overrides
+
+    def __getattr__(self, name):
+        return getattr(self._base, name)
+
+    def _patch(self, config, key: str):
+        patches = self._overrides.get(key, {})
+        for k, v in patches.items():
+            if hasattr(config, k):
+                setattr(config, k, v)
+        return config
+
+    def frame_extraction_config(self):
+        return self._patch(self._base.frame_extraction_config(), "frame_extraction")
+
+    def scene_filter_config(self):
+        return self._patch(self._base.scene_filter_config(), "scene_filter")
+
+    def detection_config(self):
+        return self._patch(self._base.detection_config(), "detection")
+
+    def ocr_config(self):
+        return self._patch(self._base.ocr_config(), "ocr")
+
+    def resolver_config(self):
+        return self._patch(self._base.resolver_config(), "resolver")
+
+    def vlm_prompt(self, crop_context):
+        return self._base.vlm_prompt(crop_context)
+
+    def aggregate(self, detections):
+        return self._base.aggregate(detections)
+
+    def auxiliary_detections(self, source):
+        return self._base.auxiliary_detections(source)
+
+
+def replay_from(
+    job_id: str,
+    start_stage: str,
+    config_overrides: dict | None = None,
+    checkpoint: bool = True,
+) -> dict:
+    """
+    Replay the pipeline from a specific stage.
+
+    Loads the checkpoint from the stage immediately before start_stage,
+    applies config overrides, and runs the subgraph from start_stage onward.
+
+    Returns the final state dict.
+    """
+    if start_stage not in NODES:
+        raise ValueError(f"Unknown stage: {start_stage!r}. Options: {NODES}")
+
+    start_idx = NODES.index(start_stage)
+
+    # Load checkpoint from the stage before start_stage
+    if start_idx == 0:
+        raise ValueError("Cannot replay from the first stage — just run the full pipeline")
+
+    previous_stage = NODES[start_idx - 1]
+
+    available = list_checkpoints(job_id)
+    if previous_stage not in available:
+        raise ValueError(
+            f"No checkpoint for stage {previous_stage!r} (job {job_id}). "
+            f"Available: {available}"
+        )
+
+    logger.info("Replaying job %s from %s (loading checkpoint: %s)",
+                job_id, start_stage, previous_stage)
+
+    state = load_checkpoint(job_id, previous_stage)
+
+    # Apply config overrides
+    if config_overrides:
+        state["config_overrides"] = config_overrides
+
+    # Set run context for SSE events
+    run_id = str(uuid.uuid4())[:8]
+    emit.set_run_context(
+        run_id=run_id,
+        parent_job_id=job_id,
+        run_type="replay",
+    )
+
+    # Build subgraph starting from start_stage
+    graph = build_graph(checkpoint=checkpoint, start_from=start_stage)
+    pipeline = graph.compile()
+
+    try:
+        result = pipeline.invoke(state)
+    finally:
+        emit.clear_run_context()
+
+    return result

detect/checkpoint/serializer.py

@@ -0,0 +1,133 @@
+"""State serialization — DetectState ↔ JSON-compatible dict."""
+
+from __future__ import annotations
+
+import dataclasses
+
+from detect.models import (
+    BoundingBox,
+    BrandDetection,
+    Frame,
+    PipelineStats,
+    TextCandidate,
+)
+
+
+# ---------------------------------------------------------------------------
+# Serialize helpers
+# ---------------------------------------------------------------------------
+
+def serialize_frame_meta(frame: Frame) -> dict:
+    meta = {
+        "sequence": frame.sequence,
+        "chunk_id": frame.chunk_id,
+        "timestamp": frame.timestamp,
+        "perceptual_hash": frame.perceptual_hash,
+    }
+    return meta
+
+
+def serialize_text_candidate(tc: TextCandidate) -> dict:
+    bbox_dict = dataclasses.asdict(tc.bbox)
+    candidate = {
+        "frame_sequence": tc.frame.sequence,
+        "bbox": bbox_dict,
+        "text": tc.text,
+        "ocr_confidence": tc.ocr_confidence,
+    }
+    return candidate
+
+
+def serialize_state(state: dict, frames_manifest: dict[int, str]) -> dict:
+    """
+    Serialize DetectState to a JSON-compatible dict.
+
+    Frame images are replaced with S3 key references.
+    TextCandidate.frame references become frame_sequence integers.
+    """
+    frames = state.get("frames", [])
+    filtered = state.get("filtered_frames", [])
+
+    manifest_strs = {str(k): v for k, v in frames_manifest.items()}
+    frames_meta = [serialize_frame_meta(f) for f in frames]
+    filtered_seqs = [f.sequence for f in filtered]
+
+    boxes_serialized = {}
+    for seq, boxes in state.get("boxes_by_frame", {}).items():
+        boxes_serialized[str(seq)] = [dataclasses.asdict(b) for b in boxes]
+
+    text_candidates = [serialize_text_candidate(tc) for tc in state.get("text_candidates", [])]
+    unresolved = [serialize_text_candidate(tc) for tc in state.get("unresolved_candidates", [])]
+    detections = [dataclasses.asdict(d) for d in state.get("detections", [])]
+    stats = dataclasses.asdict(state.get("stats", PipelineStats()))
+
+    checkpoint = {
+        "job_id": state.get("job_id", ""),
+        "video_path": state.get("video_path", ""),
+        "profile_name": state.get("profile_name", ""),
+        "config_overrides": state.get("config_overrides", {}),
+        "frames_manifest": manifest_strs,
+        "frames_meta": frames_meta,
+        "filtered_frame_sequences": filtered_seqs,
+        "boxes_by_frame": boxes_serialized,
+        "text_candidates": text_candidates,
+        "unresolved_candidates": unresolved,
+        "detections": detections,
+        "stats": stats,
+    }
+    return checkpoint
+
+
+# ---------------------------------------------------------------------------
+# Deserialize helpers
+# ---------------------------------------------------------------------------
+
+def deserialize_text_candidate(d: dict, frame_map: dict[int, Frame]) -> TextCandidate:
+    frame = frame_map[d["frame_sequence"]]
+    bbox = BoundingBox(**d["bbox"])
+    candidate = TextCandidate(
+        frame=frame,
+        bbox=bbox,
+        text=d["text"],
+        ocr_confidence=d["ocr_confidence"],
+    )
+    return candidate
+
+
+def deserialize_state(checkpoint: dict, frames: list[Frame]) -> dict:
+    """Reconstitute DetectState from a checkpoint dict + loaded frames."""
+    frame_map = {f.sequence: f for f in frames}
+
+    filtered_seqs = set(checkpoint.get("filtered_frame_sequences", []))
+    filtered_frames = [f for f in frames if f.sequence in filtered_seqs]
+
+    boxes_by_frame = {}
+    for seq_str, box_dicts in checkpoint.get("boxes_by_frame", {}).items():
+        seq = int(seq_str)
+        boxes_by_frame[seq] = [BoundingBox(**b) for b in box_dicts]
+
+    text_candidates = [
+        deserialize_text_candidate(d, frame_map)
+        for d in checkpoint.get("text_candidates", [])
+    ]
+    unresolved_candidates = [
+        deserialize_text_candidate(d, frame_map)
+        for d in checkpoint.get("unresolved_candidates", [])
+    ]
+    detections = [BrandDetection(**d) for d in checkpoint.get("detections", [])]
+    stats = PipelineStats(**checkpoint.get("stats", {}))
+
+    state = {
+        "job_id": checkpoint.get("job_id", ""),
+        "video_path": checkpoint.get("video_path", ""),
+        "profile_name": checkpoint.get("profile_name", ""),
+        "config_overrides": checkpoint.get("config_overrides", {}),
+        "frames": frames,
+        "filtered_frames": filtered_frames,
+        "boxes_by_frame": boxes_by_frame,
+        "text_candidates": text_candidates,
+        "unresolved_candidates": unresolved_candidates,
+        "detections": detections,
+        "stats": stats,
+    }
+    return state

detect/checkpoint/storage.py

@@ -0,0 +1,215 @@
+"""
+Checkpoint storage — save/load stage state.
+
+Binary data (frame images) → S3/MinIO via frames.py
+Structured data (boxes, detections, stats, config) → Postgres via Django ORM
+
+Until the Django model is generated by modelgen, checkpoint data is stored
+as JSON in S3 as a fallback. Once DetectJob/StageCheckpoint models exist,
+this module switches to Postgres.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import os
+import tempfile
+from pathlib import Path
+
+from .frames import save_frames, load_frames, BUCKET, CHECKPOINT_PREFIX
+from .serializer import serialize_state, deserialize_state
+
+logger = logging.getLogger(__name__)
+
+
+def _has_db() -> bool:
+    """Check if the DB layer is available (Django + models generated by modelgen)."""
+    try:
+        from core.db.detect import get_stage_checkpoint as _
+        # Quick check that the model exists (modelgen may not have run yet)
+        from admin.mpr.media_assets.models import StageCheckpoint as _
+        return True
+    except (ImportError, Exception):
+        return False
+
+
+# ---------------------------------------------------------------------------
+# Save
+# ---------------------------------------------------------------------------
+
+def save_checkpoint(
+    job_id: str,
+    stage: str,
+    stage_index: int,
+    state: dict,
+    frames_manifest: dict[int, str] | None = None,
+) -> str:
+    """
+    Save a stage checkpoint.
+
+    Saves frame images to S3 (if not already saved), then persists
+    structured state to Postgres (or S3 JSON fallback).
+
+    Returns the checkpoint identifier (DB id or S3 key).
+    """
+    # Save frames to S3 if no manifest provided
+    if frames_manifest is None:
+        all_frames = state.get("frames", [])
+        frames_manifest = save_frames(job_id, all_frames)
+
+    checkpoint_data = serialize_state(state, frames_manifest)
+
+    if _has_db():
+        checkpoint_id = _save_to_db(job_id, stage, stage_index, checkpoint_data)
+    else:
+        checkpoint_id = _save_to_s3(job_id, stage, checkpoint_data)
+
+    return checkpoint_id
+
+
+def _save_to_db(job_id: str, stage: str, stage_index: int, data: dict) -> str:
+    """Save checkpoint structured data to Postgres via core/db."""
+    import uuid
+    from core.db.detect import save_stage_checkpoint
+
+    job_uuid = uuid.UUID(job_id) if isinstance(job_id, str) else job_id
+    checkpoint_id = uuid.uuid4()
+    frames_prefix = f"{CHECKPOINT_PREFIX}/{job_id}/frames/"
+
+    checkpoint = save_stage_checkpoint(
+        id=checkpoint_id,
+        job_id=job_uuid,
+        stage=stage,
+        stage_index=stage_index,
+        frames_prefix=frames_prefix,
+        frames_manifest=data.get("frames_manifest", {}),
+        frames_meta=data.get("frames_meta", []),
+        filtered_frame_sequences=data.get("filtered_frame_sequences", []),
+        boxes_by_frame=data.get("boxes_by_frame", {}),
+        text_candidates=data.get("text_candidates", []),
+        unresolved_candidates=data.get("unresolved_candidates", []),
+        detections=data.get("detections", []),
+        stats=data.get("stats", {}),
+        config_snapshot=data.get("config_overrides", {}),
+        config_overrides=data.get("config_overrides", {}),
+        video_path=data.get("video_path", ""),
+        profile_name=data.get("profile_name", ""),
+    )
+
+    logger.info("Checkpoint saved to DB: %s/%s (id=%s)", job_id, stage, checkpoint.id)
+    return str(checkpoint.id)
+
+
+def _save_to_s3(job_id: str, stage: str, data: dict) -> str:
+    """Fallback: save checkpoint as JSON to S3 (before modelgen generates DB models)."""
+    from core.storage.s3 import upload_file
+
+    checkpoint_json = json.dumps(data, default=str)
+    key = f"{CHECKPOINT_PREFIX}/{job_id}/stages/{stage}.json"
+
+    with tempfile.NamedTemporaryFile(mode="w", suffix=".json", delete=False) as tmp:
+        tmp.write(checkpoint_json)
+        tmp_path = tmp.name
+
+    try:
+        upload_file(tmp_path, BUCKET, key)
+    finally:
+        os.unlink(tmp_path)
+
+    logger.info("Checkpoint saved to S3: s3://%s/%s", BUCKET, key)
+    return key
+
+
+# ---------------------------------------------------------------------------
+# Load
+# ---------------------------------------------------------------------------
+
+def load_checkpoint(job_id: str, stage: str) -> dict:
+    """
+    Load a stage checkpoint and reconstitute full DetectState.
+
+    Tries Postgres first, falls back to S3 JSON.
+    """
+    if _has_db():
+        data = _load_from_db(job_id, stage)
+    else:
+        data = _load_from_s3(job_id, stage)
+
+    raw_manifest = data.get("frames_manifest", {})
+    manifest = {int(k): v for k, v in raw_manifest.items()}
+    frame_metadata = data.get("frames_meta", [])
+    frames = load_frames(manifest, frame_metadata)
+
+    state = deserialize_state(data, frames)
+
+    logger.info("Checkpoint loaded: %s/%s (%d frames)", job_id, stage, len(frames))
+    return state
+
+
+def _load_from_db(job_id: str, stage: str) -> dict:
+    """Load checkpoint data from Postgres via core/db."""
+    from core.db.detect import get_stage_checkpoint
+
+    checkpoint = get_stage_checkpoint(job_id, stage)
+
+    data = {
+        "job_id": str(checkpoint.job_id),
+        "video_path": checkpoint.video_path,
+        "profile_name": checkpoint.profile_name,
+        "config_overrides": checkpoint.config_overrides,
+        "frames_manifest": checkpoint.frames_manifest,
+        "frames_meta": checkpoint.frames_meta,
+        "filtered_frame_sequences": checkpoint.filtered_frame_sequences,
+        "boxes_by_frame": checkpoint.boxes_by_frame,
+        "text_candidates": checkpoint.text_candidates,
+        "unresolved_candidates": checkpoint.unresolved_candidates,
+        "detections": checkpoint.detections,
+        "stats": checkpoint.stats,
+    }
+    return data
+
+
+def _load_from_s3(job_id: str, stage: str) -> dict:
+    """Fallback: load checkpoint JSON from S3."""
+    from core.storage.s3 import download_to_temp
+
+    key = f"{CHECKPOINT_PREFIX}/{job_id}/stages/{stage}.json"
+    tmp_path = download_to_temp(BUCKET, key)
+    try:
+        with open(tmp_path) as f:
+            data = json.load(f)
+    finally:
+        os.unlink(tmp_path)
+
+    return data
+
+
+# ---------------------------------------------------------------------------
+# List
+# ---------------------------------------------------------------------------
+
+def list_checkpoints(job_id: str) -> list[str]:
+    """List available checkpoint stages for a job."""
+    if _has_db():
+        return _list_from_db(job_id)
+    return _list_from_s3(job_id)
+
+
+def _list_from_db(job_id: str) -> list[str]:
+    from core.db.detect import list_stage_checkpoints
+    return list_stage_checkpoints(job_id)
+
+
+def _list_from_s3(job_id: str) -> list[str]:
+    from core.storage.s3 import list_objects
+
+    prefix = f"{CHECKPOINT_PREFIX}/{job_id}/stages/"
+    objects = list_objects(BUCKET, prefix)
+
+    stages = []
+    for obj in objects:
+        name = Path(obj["key"]).stem
+        stages.append(name)
+
+    return stages

detect/checkpoint/tasks.py

@@ -0,0 +1,71 @@
+"""
+Celery tasks for detection pipeline async operations.
+
+retry_candidates: re-run VLM/cloud escalation with different config.
+"""
+
+from __future__ import annotations
+
+import logging
+import uuid
+from datetime import datetime, timezone
+
+from celery import shared_task
+
+logger = logging.getLogger(__name__)
+
+
+@shared_task(bind=True, max_retries=1, default_retry_delay=30)
+def retry_candidates(
+    self,
+    job_id: str,
+    config_overrides: dict | None = None,
+    start_stage: str = "escalate_vlm",
+):
+    """
+    Retry unresolved candidates with different config.
+
+    Loads the checkpoint from the stage before start_stage,
+    applies config overrides (e.g. different cloud provider),
+    and runs from start_stage onward.
+    """
+    from detect.checkpoint.replay import replay_from
+
+    run_id = str(uuid.uuid4())[:8]
+    logger.info("Retry task %s: job=%s, from=%s, overrides=%s",
+                run_id, job_id, start_stage, config_overrides)
+
+    try:
+        result = replay_from(
+            job_id=job_id,
+            start_stage=start_stage,
+            config_overrides=config_overrides,
+        )
+
+        detections = result.get("detections", [])
+        report = result.get("report")
+        brands_found = len(report.brands) if report else 0
+
+        logger.info("Retry %s complete: %d detections, %d brands",
+                    run_id, len(detections), brands_found)
+
+        return {
+            "status": "completed",
+            "run_id": run_id,
+            "job_id": job_id,
+            "detections": len(detections),
+            "brands_found": brands_found,
+        }
+
+    except Exception as e:
+        logger.exception("Retry %s failed: %s", run_id, e)
+
+        if self.request.retries < self.max_retries:
+            raise self.retry(exc=e)
+
+        return {
+            "status": "failed",
+            "run_id": run_id,
+            "job_id": job_id,
+            "error": str(e),
+        }

detect/emit.py

@@ -3,6 +3,9 @@ Event emission helpers for detection pipeline stages.
 
 Single place that knows how to build event payloads.
 Stages call these instead of constructing dicts or dataclasses directly.
+
+Run context (run_id, parent_job_id) is set once at pipeline start via
+set_run_context() and automatically injected into all events.
 """
 
 from __future__ import annotations
@@ -13,9 +16,33 @@ from datetime import datetime, timezone
 from detect.events import push_detect_event
 from detect.models import PipelineStats
 
+# Module-level run context — set once per pipeline invocation
+_run_context: dict = {}
+
+
+def set_run_context(run_id: str = "", parent_job_id: str = "", run_type: str = "initial"):
+    """Set the run context for all subsequent events in this pipeline invocation."""
+    global _run_context
+    _run_context = {
+        "run_id": run_id,
+        "parent_job_id": parent_job_id,
+        "run_type": run_type,
+    }
+
+
+def clear_run_context():
+    global _run_context
+    _run_context = {}
+
+
+def _inject_context(payload: dict) -> dict:
+    """Add run context fields to an event payload."""
+    if _run_context:
+        payload.update(_run_context)
+    return payload
+
 
 def log(job_id: str | None, stage: str, level: str, msg: str) -> None:
-    """Emit a log event."""
     if not job_id:
         return
     payload = {
@@ -24,15 +51,17 @@ def log(job_id: str | None, stage: str, level: str, msg: str) -> None:
         "msg": msg,
         "ts": datetime.now(timezone.utc).isoformat(),
     }
+    _inject_context(payload)
     push_detect_event(job_id, "log", payload)
 
 
 def stats(job_id: str | None, **kwargs) -> None:
-    """Emit a stats_update event. Pass only the fields that changed."""
     if not job_id:
         return
     s = PipelineStats(**kwargs)
-    push_detect_event(job_id, "stats_update", dataclasses.asdict(s))
+    payload = dataclasses.asdict(s)
+    _inject_context(payload)
+    push_detect_event(job_id, "stats_update", payload)
 
 
 def frame_update(
@@ -42,7 +71,6 @@ def frame_update(
     jpeg_b64: str,
     boxes: list[dict],
 ) -> None:
-    """Emit a frame_update event with the image and bounding boxes."""
     if not job_id:
         return
     payload = {
@@ -51,14 +79,15 @@ def frame_update(
         "jpeg_b64": jpeg_b64,
         "boxes": boxes,
     }
+    _inject_context(payload)
     push_detect_event(job_id, "frame_update", payload)
 
 
 def graph_update(job_id: str | None, nodes: list[dict]) -> None:
-    """Emit a graph_update event with node states."""
     if not job_id:
         return
     payload = {"nodes": nodes}
+    _inject_context(payload)
     push_detect_event(job_id, "graph_update", payload)
 
 
@@ -72,7 +101,6 @@ def detection(
     content_type: str = "",
     frame_ref: int | None = None,
 ) -> None:
-    """Emit a brand detection event."""
     if not job_id:
         return
     payload = {
@@ -84,12 +112,13 @@ def detection(
         "content_type": content_type,
         "frame_ref": frame_ref,
     }
+    _inject_context(payload)
     push_detect_event(job_id, "detection", payload)
 
 
 def job_complete(job_id: str | None, report: dict) -> None:
-    """Emit a job_complete event with the final report."""
     if not job_id:
         return
     payload = {"job_id": job_id, "report": report}
+    _inject_context(payload)
     push_detect_event(job_id, "job_complete", payload)

detect/graph.py

@@ -42,9 +42,17 @@ NODES = [
 def _get_profile(state: DetectState):
     name = state.get("profile_name", "soccer_broadcast")
     if name == "soccer_broadcast":
-        return SoccerBroadcastProfile()
+        profile = SoccerBroadcastProfile()
+    else:
         raise ValueError(f"Unknown profile: {name}")
 
+    overrides = state.get("config_overrides")
+    if overrides:
+        from detect.checkpoint.replay import OverrideProfile
+        profile = OverrideProfile(profile, overrides)
+
+    return profile
+
 
 # Track node states across the pipeline run
 _node_states: dict[str, dict[str, str]] = {}
@@ -68,6 +76,18 @@ def _emit_transition(state: DetectState, node: str, status: str):
 # --- Node functions ---
 
 def node_extract_frames(state: DetectState) -> dict:
+    # Set run context for initial runs (replays set it in replay_from)
+    job_id = state.get("job_id", "")
+    if job_id and not emit._run_context:
+        emit.set_run_context(run_id=job_id, parent_job_id=job_id, run_type="initial")
+
+    # Load session brands from DB for this source
+    source_asset_id = state.get("source_asset_id")
+    if source_asset_id and not state.get("session_brands"):
+        from detect.stages.brand_resolver import build_session_dict
+        session_brands = build_session_dict(source_asset_id)
+        state["session_brands"] = session_brands
+
     _emit_transition(state, "extract_frames", "running")
 
     with trace_node(state, "extract_frames") as span:
@@ -142,13 +162,16 @@ def node_match_brands(state: DetectState) -> dict:
 
     with trace_node(state, "match_brands") as span:
         profile = _get_profile(state)
-        dictionary = profile.brand_dictionary()
         resolver_config = profile.resolver_config()
         candidates = state.get("text_candidates", [])
+        session_brands = state.get("session_brands", {})
         job_id = state.get("job_id")
+        source_asset_id = state.get("source_asset_id")
 
         matched, unresolved = resolve_brands(
-            candidates, dictionary, resolver_config,
+            candidates, resolver_config,
+            session_brands=session_brands,
+            source_asset_id=source_asset_id,
             content_type=profile.name, job_id=job_id,
         )
         span.set_output({"matched": len(matched), "unresolved": len(unresolved)})
@@ -170,6 +193,7 @@ def node_escalate_vlm(state: DetectState) -> dict:
             vlm_prompt_fn=profile.vlm_prompt,
             inference_url=INFERENCE_URL,
             content_type=profile.name,
+            source_asset_id=state.get("source_asset_id"),
             job_id=job_id,
         )
 
@@ -202,6 +226,7 @@ def node_escalate_cloud(state: DetectState) -> dict:
             vlm_prompt_fn=profile.vlm_prompt,
             stats=stats,
             content_type=profile.name,
+            source_asset_id=state.get("source_asset_id"),
             job_id=job_id,
         )
 
@@ -239,33 +264,87 @@ def node_compile_report(state: DetectState) -> dict:
     return {"report": report}
 
 
+# --- Checkpoint wrapper ---
+
+_CHECKPOINT_ENABLED = os.environ.get("MPR_CHECKPOINT", "").strip() == "1"
+_frames_manifest: dict[str, dict[int, str]] = {}  # job_id → manifest (cached per job)
+
+
+def _checkpointing_node(node_name: str, node_fn):
+    """Wrap a node function to auto-checkpoint after completion."""
+    stage_index = NODES.index(node_name)
+
+    def wrapper(state: DetectState) -> dict:
+        result = node_fn(state)
+
+        job_id = state.get("job_id", "")
+        if not job_id:
+            return result
+
+        from detect.checkpoint import save_checkpoint, save_frames
+
+        merged = {**state, **result}
+
+        # Save frames once (first checkpoint), reuse manifest after
+        manifest = _frames_manifest.get(job_id)
+        if manifest is None and node_name == "extract_frames":
+            manifest = save_frames(job_id, merged.get("frames", []))
+            _frames_manifest[job_id] = manifest
+
+        save_checkpoint(job_id, node_name, stage_index, merged, frames_manifest=manifest)
+        return result
+
+    wrapper.__name__ = node_fn.__name__
+    return wrapper
+
+
 # --- Graph construction ---
 
-def build_graph() -> StateGraph:
+NODE_FUNCTIONS = [
+    ("extract_frames", node_extract_frames),
+    ("filter_scenes", node_filter_scenes),
+    ("detect_objects", node_detect_objects),
+    ("run_ocr", node_run_ocr),
+    ("match_brands", node_match_brands),
+    ("escalate_vlm", node_escalate_vlm),
+    ("escalate_cloud", node_escalate_cloud),
+    ("compile_report", node_compile_report),
+]
+
+
+def build_graph(checkpoint: bool | None = None, start_from: str | None = None) -> StateGraph:
+    """
+    Build the pipeline graph.
+
+    checkpoint: enable auto-checkpointing (default: MPR_CHECKPOINT env var)
+    start_from: skip nodes before this stage (for replay)
+    """
+    do_checkpoint = checkpoint if checkpoint is not None else _CHECKPOINT_ENABLED
+
     graph = StateGraph(DetectState)
 
-    graph.add_node("extract_frames", node_extract_frames)
+    # Filter to start_from if replaying
-    graph.add_node("filter_scenes", node_filter_scenes)
+    node_pairs = NODE_FUNCTIONS
-    graph.add_node("detect_objects", node_detect_objects)
+    if start_from:
-    graph.add_node("run_ocr", node_run_ocr)
+        start_idx = next(i for i, (name, _) in enumerate(NODE_FUNCTIONS) if name == start_from)
-    graph.add_node("match_brands", node_match_brands)
+        node_pairs = NODE_FUNCTIONS[start_idx:]
-    graph.add_node("escalate_vlm", node_escalate_vlm)
-    graph.add_node("escalate_cloud", node_escalate_cloud)
-    graph.add_node("compile_report", node_compile_report)
 
-    graph.set_entry_point("extract_frames")
+    for name, fn in node_pairs:
-    graph.add_edge("extract_frames", "filter_scenes")
+        wrapped = _checkpointing_node(name, fn) if do_checkpoint else fn
-    graph.add_edge("filter_scenes", "detect_objects")
+        graph.add_node(name, wrapped)
-    graph.add_edge("detect_objects", "run_ocr")
+
-    graph.add_edge("run_ocr", "match_brands")
+    # Wire edges
-    graph.add_edge("match_brands", "escalate_vlm")
+    entry = node_pairs[0][0]
-    graph.add_edge("escalate_vlm", "escalate_cloud")
+    graph.set_entry_point(entry)
-    graph.add_edge("escalate_cloud", "compile_report")
+
-    graph.add_edge("compile_report", END)
+    for i in range(len(node_pairs) - 1):
+        graph.add_edge(node_pairs[i][0], node_pairs[i + 1][0])
+
+    graph.add_edge(node_pairs[-1][0], END)
 
     return graph
 
 
-def get_pipeline():
+def get_pipeline(checkpoint: bool | None = None):
     """Return a compiled, runnable pipeline."""
-    return build_graph().compile()
+    return build_graph(checkpoint=checkpoint).compile()

detect/profiles/__init__.py

@@ -1,6 +1,5 @@
 from .base import (
     ContentTypeProfile,
-    BrandDictionary,
     CropContext,
     DetectionConfig,
     FrameExtractionConfig,
@@ -12,7 +11,6 @@ from .soccer import SoccerBroadcastProfile
 
 __all__ = [
     "ContentTypeProfile",
-    "BrandDictionary",
     "CropContext",
     "DetectionConfig",
     "FrameExtractionConfig",

detect/profiles/base.py

@@ -44,12 +44,6 @@ class ResolverConfig:
     fuzzy_threshold: int = 75
 
 
-@dataclass
-class BrandDictionary:
-    """Maps canonical brand name → list of known aliases/spellings."""
-    brands: dict[str, list[str]] = field(default_factory=dict)
-
-
 @dataclass
 class CropContext:
     image: bytes
@@ -64,7 +58,6 @@ class ContentTypeProfile(Protocol):
     def scene_filter_config(self) -> SceneFilterConfig: ...
     def detection_config(self) -> DetectionConfig: ...
     def ocr_config(self) -> OCRConfig: ...
-    def brand_dictionary(self) -> BrandDictionary: ...
     def resolver_config(self) -> ResolverConfig: ...
     def vlm_prompt(self, crop_context: CropContext) -> str: ...
     def aggregate(self, detections: list[BrandDetection]) -> DetectionReport: ...

detect/profiles/soccer.py

@@ -5,7 +5,6 @@ from __future__ import annotations
 from detect.models import BrandDetection, BrandStats, DetectionReport, PipelineStats
 
 from .base import (
-    BrandDictionary,
     CropContext,
     DetectionConfig,
     FrameExtractionConfig,
@@ -34,22 +33,6 @@ class SoccerBroadcastProfile:
     def ocr_config(self) -> OCRConfig:
         return OCRConfig(languages=["en", "es"], min_confidence=0.5)
 
-    def brand_dictionary(self) -> BrandDictionary:
-        return BrandDictionary(brands={
-            "Nike": ["nike", "NIKE", "swoosh"],
-            "Adidas": ["adidas", "ADIDAS", "adi"],
-            "Puma": ["puma", "PUMA"],
-            "Emirates": ["emirates", "fly emirates", "EMIRATES"],
-            "Coca-Cola": ["coca-cola", "coca cola", "coke", "COCA-COLA"],
-            "Pepsi": ["pepsi", "PEPSI"],
-            "Mastercard": ["mastercard", "MASTERCARD"],
-            "Heineken": ["heineken", "HEINEKEN"],
-            "Santander": ["santander", "SANTANDER"],
-            "Gazprom": ["gazprom", "GAZPROM"],
-            "Qatar Airways": ["qatar airways", "QATAR AIRWAYS"],
-            "Lay's": ["lays", "lay's", "LAYS", "LAY'S"],
-        })
-
     def resolver_config(self) -> ResolverConfig:
         return ResolverConfig(fuzzy_threshold=75)
 

detect/profiles/stubs.py

@@ -5,7 +5,6 @@ from __future__ import annotations
 from detect.models import BrandDetection, DetectionReport
 
 from .base import (
-    BrandDictionary,
     CropContext,
     DetectionConfig,
     FrameExtractionConfig,
@@ -30,9 +29,6 @@ class NewsBroadcastProfile:
     def ocr_config(self) -> OCRConfig:
         raise NotImplementedError
 
-    def brand_dictionary(self) -> BrandDictionary:
-        raise NotImplementedError
-
     def resolver_config(self) -> ResolverConfig:
         raise NotImplementedError
 
@@ -61,9 +57,6 @@ class AdvertisingProfile:
     def ocr_config(self) -> OCRConfig:
         raise NotImplementedError
 
-    def brand_dictionary(self) -> BrandDictionary:
-        raise NotImplementedError
-
     def resolver_config(self) -> ResolverConfig:
         raise NotImplementedError
 
@@ -92,9 +85,6 @@ class TranscriptProfile:
     def ocr_config(self) -> OCRConfig:
         raise NotImplementedError
 
-    def brand_dictionary(self) -> BrandDictionary:
-        raise NotImplementedError
-
     def resolver_config(self) -> ResolverConfig:
         raise NotImplementedError
 

detect/stages/brand_resolver.py

@@ -1,9 +1,17 @@
 """
-Stage 5 — Brand Resolver
+Stage 5 — Brand Resolver (discovery mode)
 
-Matches OCR text against the profile's brand dictionary.
+Discovery-first brand matching. No static dictionary — all brands live in the DB.
-Uses exact matching first, then fuzzy matching (rapidfuzz) as fallback.
+
-Emits detection events for confirmed brands.
+Flow:
+  1. Check session sightings first (brands already seen in this source)
+  2. Check global known brands (accumulated across all runs)
+  3. Unresolved candidates → escalate to VLM/cloud
+  4. Confirmed brands get added to DB for future runs
+
+The resolver is an enricher, not a gatekeeper. Every OCR text candidate
+passes through — the question is whether we can resolve it cheaply (DB lookup)
+or need to escalate (VLM/cloud).
 """
 
 from __future__ import annotations
@@ -14,99 +22,199 @@ from rapidfuzz import fuzz
 
 from detect import emit
 from detect.models import BrandDetection, TextCandidate
-from detect.profiles.base import BrandDictionary, ResolverConfig
+from detect.profiles.base import ResolverConfig
 
 logger = logging.getLogger(__name__)
 
 
 def _normalize(text: str) -> str:
-    """Normalize text for matching."""
     return text.strip().lower()
 
 
-def _exact_match(text: str, dictionary: BrandDictionary) -> str | None:
+def _has_db() -> bool:
-    """Try exact match against all aliases."""
+    try:
+        from core.db.detect import find_brand_by_text as _
+        from admin.mpr.media_assets.models import KnownBrand as _
+        return True
+    except (ImportError, Exception):
+        return False
+
+
+def _match_session(text: str, session_brands: dict[str, str]) -> str | None:
+    """
+    Check against session brands (already seen in this source).
+
+    session_brands: {normalized_name: canonical_name, ...}
+    Includes aliases.
+    """
     normalized = _normalize(text)
-    for canonical, aliases in dictionary.brands.items():
+    return session_brands.get(normalized)
-        if normalized == _normalize(canonical):
-            return canonical
-        for alias in aliases:
-            if normalized == _normalize(alias):
-                return canonical
-    return None
 
 
-def _fuzzy_match(text: str, dictionary: BrandDictionary, threshold: int) -> tuple[str | None, int]:
+def _match_known(text: str, threshold: int) -> tuple[str | None, str | None]:
-    """Try fuzzy match, return (brand, score) or (None, 0)."""
+    """
+    Check against global known brands in DB.
+
+    Returns (canonical_name, brand_id) or (None, None).
+    """
+    if not _has_db():
+        return None, None
+
+    from core.db.detect import find_brand_by_text
+    brand = find_brand_by_text(text)
+    if brand:
+        return brand.canonical_name, str(brand.id)
+
+    # Fuzzy match against all known brands
+    from core.db.detect import list_all_brands
+    all_brands = list_all_brands()
+
     normalized = _normalize(text)
     best_brand = None
     best_score = 0
 
-    for canonical, aliases in dictionary.brands.items():
+    for known in all_brands:
-        all_names = [canonical] + aliases
+        names = [known.canonical_name] + (known.aliases or [])
-        for name in all_names:
+        for name in names:
             score = fuzz.ratio(normalized, _normalize(name))
             if score > best_score and score >= threshold:
                 best_score = score
-                best_brand = canonical
+                best_brand = known
 
-    return best_brand, best_score
+    if best_brand:
+        return best_brand.canonical_name, str(best_brand.id)
+
+    return None, None
+
+
+def _register_brand(canonical_name: str, source: str) -> str | None:
+    """Register a newly discovered brand in the DB. Returns brand_id."""
+    if not _has_db():
+        return None
+
+    from core.db.detect import get_or_create_brand
+    brand, created = get_or_create_brand(canonical_name, source=source)
+    if created:
+        logger.info("New brand discovered: %s (source=%s)", canonical_name, source)
+    return str(brand.id)
+
+
+def _record_sighting(source_asset_id: str | None, brand_id: str,
+                     brand_name: str, timestamp: float,
+                     confidence: float, source: str):
+    """Record a brand sighting for this source."""
+    if not _has_db() or not source_asset_id:
+        return
+
+    from core.db.detect import record_sighting
+    import uuid
+    asset_id = uuid.UUID(source_asset_id) if isinstance(source_asset_id, str) else source_asset_id
+    brand_uuid = uuid.UUID(brand_id) if isinstance(brand_id, str) else brand_id
+    record_sighting(asset_id, brand_uuid, brand_name, timestamp, confidence, source)
+
+
+def build_session_dict(source_asset_id: str | None) -> dict[str, str]:
+    """
+    Load session brands from DB for this source.
+
+    Returns {normalized_name: canonical_name, ...} including aliases.
+    """
+    if not _has_db() or not source_asset_id:
+        return {}
+
+    from core.db.detect import get_source_sightings
+    import uuid
+
+    asset_id = uuid.UUID(source_asset_id) if isinstance(source_asset_id, str) else source_asset_id
+    sightings = get_source_sightings(asset_id)
+
+    session = {}
+    for s in sightings:
+        canonical = s.brand_name
+        session[_normalize(canonical)] = canonical
+
+    # Also load aliases from KnownBrand for each sighted brand
+    if _has_db():
+        from core.db.detect import list_all_brands
+        all_brands = list_all_brands()
+        sighted_names = {s.brand_name for s in sightings}
+        for brand in all_brands:
+            if brand.canonical_name in sighted_names:
+                for alias in (brand.aliases or []):
+                    session[_normalize(alias)] = brand.canonical_name
+
+    return session
 
 
 def resolve_brands(
     candidates: list[TextCandidate],
-    dictionary: BrandDictionary,
     config: ResolverConfig,
+    session_brands: dict[str, str] | None = None,
+    source_asset_id: str | None = None,
     content_type: str = "",
     job_id: str | None = None,
 ) -> tuple[list[BrandDetection], list[TextCandidate]]:
     """
-    Match text candidates against the brand dictionary.
+    Match text candidates against known brands (session → global → unresolved).
 
-    Returns:
+    session_brands: pre-loaded session dict (from build_session_dict)
-        - matched: list of BrandDetection for confirmed brands
+    source_asset_id: for recording new sightings in DB
-        - unresolved: list of TextCandidate that couldn't be matched
     """
+    if session_brands is None:
+        session_brands = {}
+
     emit.log(job_id, "BrandResolver", "INFO",
-             f"Matching {len(candidates)} candidates against "
+             f"Resolving {len(candidates)} candidates "
-             f"{len(dictionary.brands)} brands (fuzzy_threshold={config.fuzzy_threshold})")
+             f"(session={len(session_brands)} brands, fuzzy={config.fuzzy_threshold})")
 
     matched: list[BrandDetection] = []
     unresolved: list[TextCandidate] = []
-    exact_count = 0
+    session_hits = 0
-    fuzzy_count = 0
+    known_hits = 0
 
     for candidate in candidates:
-        # Try exact match first
+        text = candidate.text
-        brand = _exact_match(candidate.text, dictionary)
+        brand_name = None
-        source = "ocr"
+        brand_id = None
+        match_source = "ocr"
 
-        if brand:
+        # 1. Check session (cheapest — in-memory dict)
-            exact_count += 1
+        brand_name = _match_session(text, session_brands)
+        if brand_name:
+            session_hits += 1
         else:
-            # Try fuzzy match
+            # 2. Check global known brands (DB query + fuzzy)
-            brand, score = _fuzzy_match(candidate.text, dictionary, config.fuzzy_threshold)
+            brand_name, brand_id = _match_known(text, config.fuzzy_threshold)
-            if brand:
+            if brand_name:
-                fuzzy_count += 1
+                known_hits += 1
+                # Add to session for subsequent candidates in this run
+                session_brands[_normalize(brand_name)] = brand_name
 
-        if brand:
+        if brand_name:
             detection = BrandDetection(
-                brand=brand,
+                brand=brand_name,
                 timestamp=candidate.frame.timestamp,
                 duration=0.5,
                 confidence=candidate.ocr_confidence,
-                source=source,
+                source=match_source,
                 bbox=candidate.bbox,
                 frame_ref=candidate.frame.sequence,
                 content_type=content_type,
             )
             matched.append(detection)
 
+            # Record sighting in DB
+            if brand_id:
+                _record_sighting(
+                    source_asset_id, brand_id, brand_name,
+                    candidate.frame.timestamp, candidate.ocr_confidence, match_source,
+                )
+
             emit.detection(
                 job_id,
-                brand=brand,
+                brand=brand_name,
                 confidence=candidate.ocr_confidence,
-                source=source,
+                source=match_source,
                 timestamp=candidate.frame.timestamp,
                 content_type=content_type,
                 frame_ref=candidate.frame.sequence,
@@ -115,7 +223,7 @@ def resolve_brands(
             unresolved.append(candidate)
 
     emit.log(job_id, "BrandResolver", "INFO",
-             f"Exact: {exact_count}, Fuzzy: {fuzzy_count}, "
+             f"Session: {session_hits}, Known: {known_hits}, "
-             f"Unresolved: {len(unresolved)} → escalating to VLM")
+             f"Unresolved: {len(unresolved)} → escalating")
 
     return matched, unresolved

detect/stages/vlm_cloud.py

@@ -27,6 +27,18 @@ logger = logging.getLogger(__name__)
 ESTIMATED_TOKENS_PER_CROP = 500
 
 
+def _register_discovered_brand(brand: str, source_asset_id: str | None,
+                               timestamp: float, confidence: float):
+    """Register a cloud-confirmed brand in the DB."""
+    try:
+        from detect.stages.brand_resolver import _register_brand, _record_sighting
+        brand_id = _register_brand(brand, "cloud_llm")
+        if brand_id and source_asset_id:
+            _record_sighting(source_asset_id, brand_id, brand, timestamp, confidence, "cloud_llm")
+    except Exception as e:
+        logger.debug("Failed to register brand %s: %s", brand, e)
+
+
 def _encode_crop(crop: np.ndarray) -> str:
     img = Image.fromarray(crop)
     buf = io.BytesIO()
@@ -84,6 +96,7 @@ def escalate_cloud(
     stats: PipelineStats,
     min_confidence: float = 0.4,
     content_type: str = "",
+    source_asset_id: str | None = None,
     job_id: str | None = None,
 ) -> list[BrandDetection]:
     """
@@ -158,6 +171,10 @@ def escalate_cloud(
                 frame_ref=candidate.frame.sequence,
             )
 
+            # Register newly discovered brand in DB
+            _register_discovered_brand(brand, source_asset_id,
+                                       candidate.frame.timestamp, confidence)
+
     stats.estimated_cloud_cost_usd += total_cost
     stats.regions_escalated_to_cloud_llm = len(candidates)
 

detect/stages/vlm_local.py

@@ -19,6 +19,18 @@ from detect.profiles.base import CropContext
 logger = logging.getLogger(__name__)
 
 
+def _register_discovered_brand(brand: str, source_asset_id: str | None,
+                               timestamp: float, confidence: float, source: str):
+    """Register a VLM-confirmed brand in the DB."""
+    try:
+        from detect.stages.brand_resolver import _register_brand, _record_sighting
+        brand_id = _register_brand(brand, source)
+        if brand_id and source_asset_id:
+            _record_sighting(source_asset_id, brand_id, brand, timestamp, confidence, source)
+    except Exception as e:
+        logger.debug("Failed to register brand %s: %s", brand, e)
+
+
 def _crop_image(candidate: TextCandidate) -> np.ndarray:
     frame = candidate.frame
     box = candidate.bbox
@@ -36,6 +48,7 @@ def escalate_vlm(
     inference_url: str | None = None,
     min_confidence: float = 0.5,
     content_type: str = "",
+    source_asset_id: str | None = None,
     job_id: str | None = None,
 ) -> tuple[list[BrandDetection], list[TextCandidate]]:
     """
@@ -107,6 +120,10 @@ def escalate_vlm(
                 frame_ref=candidate.frame.sequence,
             )
 
+            # Register newly discovered brand in DB
+            _register_discovered_brand(brand, source_asset_id,
+                                       candidate.frame.timestamp, confidence, "local_vlm")
+
             logger.debug("VLM matched: %s (%.2f) — %s", brand, confidence, reasoning)
         else:
             still_unresolved.append(candidate)

detect/state.py

@@ -17,6 +17,7 @@ class DetectState(TypedDict, total=False):
     video_path: str
     job_id: str
     profile_name: str
+    source_asset_id: str  # UUID of the source MediaAsset
 
     # Stage outputs
     frames: list[Frame]
@@ -27,5 +28,11 @@ class DetectState(TypedDict, total=False):
     detections: list[BrandDetection]
     report: DetectionReport
 
+    # Session brands (accumulated during the run, persisted to DB)
+    session_brands: dict  # {normalized_name: canonical_name}
+
     # Running stats (updated by each stage)
     stats: PipelineStats
+
+    # Config overrides for replay (applied via OverrideProfile)
+    config_overrides: dict

tests/detect/manual/test_escalation_e2e.py

@@ -31,8 +31,16 @@ def ts():
     return datetime.now(timezone.utc).isoformat()
 
 
+RUN_CONTEXT = {}
+
+
+def set_run_context(run_id: str, parent_job_id: str, run_type: str = "initial"):
+    RUN_CONTEXT.update({"run_id": run_id, "parent_job_id": parent_job_id, "run_type": run_type})
+
+
 def push(r, key, event):
     event["ts"] = event.get("ts", ts())
+    event.update(RUN_CONTEXT)
     r.rpush(key, json.dumps(event))
     return event
 
@@ -85,7 +93,11 @@ def main():
     r.delete(key)
     delay = args.delay
 
+    run_id = f"{args.job[:8]}-r1"
+    set_run_context(run_id=run_id, parent_job_id=args.job, run_type="initial")
+
     logger.info("Full escalation pipeline simulation → %s", key)
+    logger.info("Run: %s (parent: %s)", run_id, args.job)
     logger.info("Open: http://mpr.local.ar/detection/?job=%s", args.job)
     input("\nPress Enter to start...")
 

tests/detect/manual/test_replay.py

@@ -0,0 +1,123 @@
+#!/usr/bin/env python3
+"""
+Test checkpoint + replay flow end-to-end.
+
+1. Runs the pipeline with checkpointing enabled on a test video
+2. Lists available checkpoints
+3. Replays from run_ocr with different config
+4. Compares detection counts
+
+Usage:
+    MPR_CHECKPOINT=1 INFERENCE_URL=http://mcrndeb:8000 python tests/detect/manual/test_replay.py [--job JOB_ID]
+
+Requires: inference server running, MinIO/S3 running, test video available
+"""
+
+import argparse
+import logging
+import os
+import sys
+from pathlib import Path
+
+# Load ctrl/.env
+env_file = Path(__file__).resolve().parents[3] / "ctrl" / ".env"
+if env_file.exists():
+    for line in env_file.read_text().splitlines():
+        line = line.strip()
+        if line and not line.startswith("#") and "=" in line:
+            key, _, value = line.partition("=")
+            os.environ.setdefault(key.strip(), value.strip())
+
+sys.path.insert(0, ".")
+
+logging.basicConfig(level=logging.INFO, format="%(levelname)-7s %(name)s — %(message)s")
+logger = logging.getLogger(__name__)
+
+# Force checkpointing on
+os.environ["MPR_CHECKPOINT"] = "1"
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    import time
+    default_job = f"replay-{int(time.time()) % 100000}"
+    parser.add_argument("--job", default=default_job)
+    parser.add_argument("--port", type=int, default=6382)
+    args = parser.parse_args()
+
+    # Override Redis to localhost (ctrl/.env has k8s hostname)
+    os.environ["REDIS_URL"] = f"redis://localhost:{args.port}/0"
+
+    from detect.graph import get_pipeline, NODES
+    from detect.checkpoint import list_checkpoints
+    from detect.checkpoint import replay_from
+    from detect.state import DetectState
+
+    VIDEO = "media/out/chunks/95043d50-4df6-4ac8-bbd5-2ba873117c6e/chunk_0000.mp4"
+
+    logger.info("Job: %s", args.job)
+    logger.info("Checkpoint: enabled")
+    logger.info("Video: %s", VIDEO)
+    logger.info("Open: http://mpr.local.ar/detection/?job=%s", args.job)
+    input("\nPress Enter to run initial pipeline...")
+
+    # --- Initial run ---
+    pipeline = get_pipeline(checkpoint=True)
+    initial_state = DetectState(
+        video_path=VIDEO,
+        job_id=args.job,
+        profile_name="soccer_broadcast",
+    )
+
+    logger.info("Running initial pipeline...")
+    result = pipeline.invoke(initial_state)
+
+    detections = result.get("detections", [])
+    report = result.get("report")
+    logger.info("Initial run: %d detections, %d brands",
+                len(detections), len(report.brands) if report else 0)
+
+    # --- List checkpoints ---
+    stages = list_checkpoints(args.job)
+    logger.info("Available checkpoints: %s", stages)
+
+    if "detect_objects" not in stages:
+        logger.error("Expected checkpoint for detect_objects — aborting replay test")
+        return
+
+    input("\nPress Enter to replay from run_ocr with different config...")
+
+    # --- Replay with different OCR config ---
+    overrides = {"ocr": {"min_confidence": 0.3, "languages": ["en", "es"]}}
+    logger.info("Replaying from run_ocr with overrides: %s", overrides)
+
+    replay_result = replay_from(
+        job_id=args.job,
+        start_stage="run_ocr",
+        config_overrides=overrides,
+    )
+
+    replay_detections = replay_result.get("detections", [])
+    replay_report = replay_result.get("report")
+    logger.info("Replay run: %d detections, %d brands",
+                len(replay_detections),
+                len(replay_report.brands) if replay_report else 0)
+
+    # --- Compare ---
+    logger.info("--- Comparison ---")
+    logger.info("Initial: %d detections", len(detections))
+    logger.info("Replay:  %d detections (min_confidence 0.5 → 0.3)", len(replay_detections))
+
+    diff = len(replay_detections) - len(detections)
+    if diff > 0:
+        logger.info("Replay found %d more detections with lower threshold", diff)
+    elif diff == 0:
+        logger.info("Same count — threshold change didn't affect this video")
+    else:
+        logger.warning("Replay found fewer detections — unexpected")
+
+    logger.info("Done.")
+
+
+if __name__ == "__main__":
+    main()

tests/detect/test_brand_resolver.py

@@ -1,20 +1,13 @@
-"""Tests for BrandResolver stage."""
+"""Tests for BrandResolver stage (discovery mode)."""
 
 import numpy as np
 import pytest
 
 from detect.models import BoundingBox, Frame, TextCandidate
-from detect.profiles.base import BrandDictionary, ResolverConfig
+from detect.profiles.base import ResolverConfig
-from detect.stages.brand_resolver import resolve_brands, _exact_match, _fuzzy_match
+from detect.stages.brand_resolver import resolve_brands, _normalize, _match_session
 
 
-DICTIONARY = BrandDictionary(brands={
-    "Nike": ["nike", "NIKE", "swoosh"],
-    "Adidas": ["adidas", "ADIDAS"],
-    "Coca-Cola": ["coca-cola", "coca cola", "coke", "COCA-COLA"],
-    "Emirates": ["emirates", "fly emirates", "EMIRATES"],
-})
-
 CONFIG = ResolverConfig(fuzzy_threshold=75)
 
 
@@ -25,57 +18,76 @@ def _make_candidate(text: str, confidence: float = 0.9) -> TextCandidate:
     return TextCandidate(frame=dummy_frame, bbox=dummy_box, text=text, ocr_confidence=confidence)
 
 
-def test_exact_match():
+def test_session_match():
-    assert _exact_match("Nike", DICTIONARY) == "Nike"
+    session = {"nike": "Nike", "fly emirates": "Emirates"}
-    assert _exact_match("nike", DICTIONARY) == "Nike"
+    assert _match_session("Nike", session) == "Nike"
-    assert _exact_match("COCA-COLA", DICTIONARY) == "Coca-Cola"
+    assert _match_session("nike", session) == "Nike"
-    assert _exact_match("fly emirates", DICTIONARY) == "Emirates"
+    assert _match_session("FLY EMIRATES", session) == "Emirates"
-    assert _exact_match("unknown brand", DICTIONARY) is None
+    assert _match_session("unknown", session) is None
 
 
-def test_fuzzy_match():
+def test_resolve_with_session(monkeypatch):
-    brand, score = _fuzzy_match("Nik3", DICTIONARY, threshold=75)
+    events = []
-    assert brand == "Nike"
+    monkeypatch.setattr("detect.emit.push_detect_event",
-    assert score >= 75
+                        lambda job_id, etype, data: events.append((etype, data)))
 
-    brand, score = _fuzzy_match("adldas", DICTIONARY, threshold=75)
+    session = {"nike": "Nike", "emirates": "Emirates"}
-    assert brand == "Adidas"
-
-    brand, score = _fuzzy_match("xyzxyzxyz", DICTIONARY, threshold=75)
-    assert brand is None
-
-
-def test_resolve_exact():
     candidates = [_make_candidate("Nike"), _make_candidate("EMIRATES")]
-    matched, unresolved = resolve_brands(candidates, DICTIONARY, CONFIG)
+
+    matched, unresolved = resolve_brands(
+        candidates, CONFIG, session_brands=session,
+    )
+
     assert len(matched) == 2
     assert len(unresolved) == 0
     assert matched[0].brand == "Nike"
     assert matched[1].brand == "Emirates"
 
 
-def test_resolve_fuzzy():
+def test_resolve_unresolved_without_db(monkeypatch):
-    candidates = [_make_candidate("coca coIa")]  # OCR misread
+    events = []
-    matched, unresolved = resolve_brands(candidates, DICTIONARY, CONFIG)
+    monkeypatch.setattr("detect.emit.push_detect_event",
-    assert len(matched) == 1
+                        lambda job_id, etype, data: events.append((etype, data)))
-    assert matched[0].brand == "Coca-Cola"
 
-
-def test_resolve_unresolved():
     candidates = [_make_candidate("random garbage text")]
-    matched, unresolved = resolve_brands(candidates, DICTIONARY, CONFIG)
+
+    matched, unresolved = resolve_brands(
+        candidates, CONFIG, session_brands={},
+    )
+
     assert len(matched) == 0
     assert len(unresolved) == 1
 
 
-def test_resolve_mixed():
+def test_resolve_empty(monkeypatch):
+    events = []
+    monkeypatch.setattr("detect.emit.push_detect_event",
+                        lambda job_id, etype, data: events.append((etype, data)))
+
+    matched, unresolved = resolve_brands([], CONFIG, session_brands={})
+
+    assert len(matched) == 0
+    assert len(unresolved) == 0
+
+
+def test_resolve_builds_session_during_run(monkeypatch):
+    """Session brands accumulate during a single run — second candidate benefits."""
+    events = []
+    monkeypatch.setattr("detect.emit.push_detect_event",
+                        lambda job_id, etype, data: events.append((etype, data)))
+
+    session = {"nike": "Nike"}
     candidates = [
-        _make_candidate("Nike"),
+        _make_candidate("Nike"),      # hits session
-        _make_candidate("unknown"),
+        _make_candidate("unknown"),   # misses everything
-        _make_candidate("adldas"),
     ]
-    matched, unresolved = resolve_brands(candidates, DICTIONARY, CONFIG)
+
-    assert len(matched) == 2  # Nike exact + Adidas fuzzy
+    matched, unresolved = resolve_brands(
+        candidates, CONFIG, session_brands=session,
+    )
+
+    assert len(matched) == 1
+    assert matched[0].brand == "Nike"
     assert len(unresolved) == 1
 
 
@@ -84,8 +96,10 @@ def test_events_emitted(monkeypatch):
     monkeypatch.setattr("detect.emit.push_detect_event",
                         lambda job_id, etype, data: events.append((etype, data)))
 
+    session = {"nike": "Nike"}
     candidates = [_make_candidate("Nike")]
-    resolve_brands(candidates, DICTIONARY, CONFIG, job_id="test-job")
+
+    resolve_brands(candidates, CONFIG, session_brands=session, job_id="test-job")
 
     event_types = [e[0] for e in events]
     assert "log" in event_types

tests/detect/test_checkpoint.py

@@ -0,0 +1,182 @@
+"""Tests for checkpoint serialization — round-trip without S3."""
+
+import numpy as np
+import pytest
+
+from detect.models import BoundingBox, BrandDetection, Frame, PipelineStats, TextCandidate
+from detect.checkpoint.serializer import (
+    serialize_state,
+    deserialize_state,
+    serialize_frame_meta,
+    serialize_text_candidate,
+    deserialize_text_candidate,
+)
+
+
+def _make_frame(seq: int = 0, w: int = 100, h: int = 80) -> Frame:
+    image = np.random.randint(0, 255, (h, w, 3), dtype=np.uint8)
+    return Frame(
+        sequence=seq,
+        chunk_id=0,
+        timestamp=float(seq) * 0.5,
+        image=image,
+        perceptual_hash=f"hash_{seq}",
+    )
+
+
+def _make_box(x=10, y=10, w=30, h=20) -> BoundingBox:
+    return BoundingBox(x=x, y=y, w=w, h=h, confidence=0.9, label="text")
+
+
+def _make_candidate(frame: Frame, text: str = "NIKE") -> TextCandidate:
+    box = _make_box()
+    return TextCandidate(frame=frame, bbox=box, text=text, ocr_confidence=0.85)
+
+
+def _make_detection(brand: str = "Nike", timestamp: float = 1.0) -> BrandDetection:
+    return BrandDetection(
+        brand=brand,
+        timestamp=timestamp,
+        duration=0.5,
+        confidence=0.92,
+        source="ocr",
+        content_type="soccer_broadcast",
+        frame_ref=0,
+    )
+
+
+# --- Frame metadata ---
+
+def test_serialize_frame_meta():
+    frame = _make_frame(seq=5)
+    meta = serialize_frame_meta(frame)
+
+    assert meta["sequence"] == 5
+    assert meta["timestamp"] == 2.5
+    assert meta["perceptual_hash"] == "hash_5"
+    assert "image" not in meta
+
+
+# --- TextCandidate ---
+
+def test_serialize_text_candidate():
+    frame = _make_frame()
+    candidate = _make_candidate(frame, text="EMIRATES")
+    data = serialize_text_candidate(candidate)
+
+    assert data["frame_sequence"] == 0
+    assert data["text"] == "EMIRATES"
+    assert data["ocr_confidence"] == 0.85
+    assert "bbox" in data
+
+
+def test_deserialize_text_candidate():
+    frame = _make_frame()
+    candidate = _make_candidate(frame, text="ADIDAS")
+
+    serialized = serialize_text_candidate(candidate)
+    frame_map = {frame.sequence: frame}
+
+    restored = deserialize_text_candidate(serialized, frame_map)
+
+    assert restored.text == "ADIDAS"
+    assert restored.ocr_confidence == 0.85
+    assert restored.frame is frame  # same object reference
+    assert restored.bbox.x == 10
+
+
+# --- Full state round-trip ---
+
+def test_state_round_trip():
+    frames = [_make_frame(seq=i) for i in range(3)]
+    filtered = frames[:2]
+
+    box = _make_box()
+    boxes_by_frame = {0: [box], 1: [box]}
+
+    candidates = [_make_candidate(frames[0], "NIKE"), _make_candidate(frames[1], "EMIRATES")]
+    unresolved = [_make_candidate(frames[2], "unknown")]
+    detections = [_make_detection("Nike", 0.5), _make_detection("Emirates", 1.0)]
+
+    stats = PipelineStats(
+        frames_extracted=3,
+        frames_after_scene_filter=2,
+        regions_detected=2,
+        regions_resolved_by_ocr=2,
+        cloud_llm_calls=1,
+        estimated_cloud_cost_usd=0.003,
+    )
+
+    state = {
+        "job_id": "test-123",
+        "video_path": "/tmp/test.mp4",
+        "profile_name": "soccer_broadcast",
+        "config_overrides": {"ocr": {"min_confidence": 0.3}},
+        "frames": frames,
+        "filtered_frames": filtered,
+        "boxes_by_frame": boxes_by_frame,
+        "text_candidates": candidates,
+        "unresolved_candidates": unresolved,
+        "detections": detections,
+        "stats": stats,
+    }
+
+    manifest = {f.sequence: f"s3://fake/frames/{f.sequence}.jpg" for f in frames}
+
+    # Serialize
+    serialized = serialize_state(state, manifest)
+
+    # Verify JSON-compatible (no numpy, no Frame objects)
+    import json
+    json_str = json.dumps(serialized, default=str)
+    assert len(json_str) > 0
+
+    # Deserialize with the original frames (simulating frame load from S3)
+    restored = deserialize_state(serialized, frames)
+
+    # Verify round-trip
+    assert restored["job_id"] == "test-123"
+    assert restored["video_path"] == "/tmp/test.mp4"
+    assert restored["profile_name"] == "soccer_broadcast"
+    assert restored["config_overrides"] == {"ocr": {"min_confidence": 0.3}}
+
+    assert len(restored["frames"]) == 3
+    assert len(restored["filtered_frames"]) == 2
+    assert len(restored["boxes_by_frame"]) == 2
+    assert len(restored["text_candidates"]) == 2
+    assert len(restored["unresolved_candidates"]) == 1
+    assert len(restored["detections"]) == 2
+
+    restored_stats = restored["stats"]
+    assert restored_stats.frames_extracted == 3
+    assert restored_stats.cloud_llm_calls == 1
+    assert restored_stats.estimated_cloud_cost_usd == 0.003
+
+    # TextCandidate frame references should point to actual Frame objects
+    tc = restored["text_candidates"][0]
+    assert tc.frame is frames[0]
+    assert tc.text == "NIKE"
+
+
+def test_state_round_trip_empty():
+    """Empty state should serialize/deserialize cleanly."""
+    state = {
+        "job_id": "empty-job",
+        "video_path": "",
+        "profile_name": "soccer_broadcast",
+        "frames": [],
+        "filtered_frames": [],
+        "boxes_by_frame": {},
+        "text_candidates": [],
+        "unresolved_candidates": [],
+        "detections": [],
+        "stats": PipelineStats(),
+    }
+
+    serialized = serialize_state(state, {})
+    restored = deserialize_state(serialized, [])
+
+    assert restored["job_id"] == "empty-job"
+    assert len(restored["frames"]) == 0
+    assert len(restored["detections"]) == 0
+    assert restored["stats"].frames_extracted == 0

tests/detect/test_profiles.py

@@ -25,11 +25,9 @@ def test_soccer_detection_config():
     assert isinstance(cfg.target_classes, list)
 
 
-def test_soccer_brand_dictionary_non_empty():
+def test_soccer_resolver_config():
-    bd = SoccerBroadcastProfile().brand_dictionary()
+    cfg = SoccerBroadcastProfile().resolver_config()
-    assert len(bd.brands) > 0
+    assert cfg.fuzzy_threshold > 0
-    for canonical, aliases in bd.brands.items():
-        assert len(aliases) > 0
 
 
 def test_soccer_vlm_prompt():
@@ -70,4 +68,4 @@ def test_stubs_raise(stub_cls):
     with pytest.raises(NotImplementedError):
         stub.frame_extraction_config()
     with pytest.raises(NotImplementedError):
-        stub.brand_dictionary()
+        stub.resolver_config()

tests/detect/test_replay.py

@@ -0,0 +1,67 @@
+"""Tests for replay and OverrideProfile."""
+
+import pytest
+
+from detect.profiles.soccer import SoccerBroadcastProfile
+from detect.checkpoint.replay import OverrideProfile
+
+
+def test_override_profile_patches_ocr():
+    base = SoccerBroadcastProfile()
+    overrides = {"ocr": {"min_confidence": 0.3, "languages": ["en", "es", "pt"]}}
+    profile = OverrideProfile(base, overrides)
+
+    config = profile.ocr_config()
+
+    assert config.min_confidence == 0.3
+    assert config.languages == ["en", "es", "pt"]
+
+
+def test_override_profile_patches_resolver():
+    base = SoccerBroadcastProfile()
+    overrides = {"resolver": {"fuzzy_threshold": 60}}
+    profile = OverrideProfile(base, overrides)
+
+    config = profile.resolver_config()
+
+    assert config.fuzzy_threshold == 60
+
+
+def test_override_profile_patches_detection():
+    base = SoccerBroadcastProfile()
+    overrides = {"detection": {"confidence_threshold": 0.5}}
+    profile = OverrideProfile(base, overrides)
+
+    config = profile.detection_config()
+
+    assert config.confidence_threshold == 0.5
+
+
+def test_override_profile_no_overrides():
+    base = SoccerBroadcastProfile()
+    profile = OverrideProfile(base, {})
+
+    ocr = profile.ocr_config()
+    base_ocr = base.ocr_config()
+
+    assert ocr.min_confidence == base_ocr.min_confidence
+    assert ocr.languages == base_ocr.languages
+
+
+def test_override_profile_delegates_non_config():
+    base = SoccerBroadcastProfile()
+    profile = OverrideProfile(base, {"ocr": {"min_confidence": 0.1}})
+
+    assert profile.name == "soccer_broadcast"
+    assert profile.resolver_config().fuzzy_threshold > 0
+
+
+def test_override_profile_ignores_unknown_fields():
+    base = SoccerBroadcastProfile()
+    overrides = {"ocr": {"nonexistent_field": 42}}
+    profile = OverrideProfile(base, overrides)
+
+    config = profile.ocr_config()
+
+    assert not hasattr(config, "nonexistent_field")
+    assert config.min_confidence == base.ocr_config().min_confidence

ui/detection-app/src/App.vue

@@ -1,6 +1,6 @@
 <script setup lang="ts">
 import { ref } from 'vue'
-import { SSEDataSource, Panel, LayoutGrid } from 'mpr-ui-framework'
+import { SSEDataSource, Panel, ResizeHandle } from 'mpr-ui-framework'
 import 'mpr-ui-framework/src/tokens.css'
 import LogPanel from './panels/LogPanel.vue'
 import FunnelPanel from './panels/FunnelPanel.vue'
@@ -9,10 +9,11 @@ import FramePanel from './panels/FramePanel.vue'
 import BrandTablePanel from './panels/BrandTablePanel.vue'
 import TimelinePanel from './panels/TimelinePanel.vue'
 import CostStatsPanel from './panels/CostStatsPanel.vue'
-import type { StatsUpdate } from './types/sse-contract'
+import type { StatsUpdate, RunContext } from './types/sse-contract'
 
 const jobId = ref(new URLSearchParams(window.location.search).get('job') || 'test-job')
 const stats = ref<StatsUpdate | null>(null)
+const runContext = ref<RunContext | null>(null)
 const status = ref<'idle' | 'live' | 'processing' | 'error'>('idle')
 
 const source = new SSEDataSource({
@@ -23,8 +24,41 @@ const source = new SSEDataSource({
 
 source.on<StatsUpdate>('stats_update', (e) => {
   stats.value = e
+  // Capture run context from first event that carries it
+  if (!runContext.value && e.run_id) {
+    runContext.value = {
+      run_id: (e as any).run_id,
+      parent_job_id: (e as any).parent_job_id,
+      run_type: (e as any).run_type ?? 'initial',
+    }
+  }
 })
 
+// Resizable splits
+const pipelineWidth = ref(320)
+const detectionsFlex = ref(3)  // ratio for detections vs stats
+const viewerHeight = ref(240)
+const timelineFlex = ref(1)
+const tableFlex = ref(1)
+
+function onPipelineResize(delta: number) {
+  pipelineWidth.value = Math.max(200, Math.min(500, pipelineWidth.value + delta))
+}
+
+function onViewerResize(delta: number) {
+  viewerHeight.value = Math.max(120, Math.min(400, viewerHeight.value + delta))
+}
+
+function onDetectionsResize(delta: number) {
+  detectionsFlex.value = Math.max(1, Math.min(6, detectionsFlex.value + delta * 0.01))
+}
+
+function onTimelineResize(delta: number) {
+  const shift = delta * 0.02
+  timelineFlex.value = Math.max(0.3, Math.min(3, timelineFlex.value + shift))
+  tableFlex.value = Math.max(0.3, Math.min(3, tableFlex.value - shift))
+}
+
 const statusMap: Record<string, 'idle' | 'live' | 'processing' | 'error'> = {
   idle: 'idle',
   connecting: 'processing',
@@ -42,41 +76,70 @@ source.connect()
     <header>
       <h1>Detection Pipeline</h1>
       <span class="status-badge" :class="status">{{ status }}</span>
+      <span v-if="runContext" class="run-info">
+        {{ runContext.run_type }} · run: {{ runContext.run_id }}
+      </span>
       <span class="job-id">job: {{ jobId }}</span>
     </header>
 
-    <LayoutGrid :columns="3" :rows="3" gap="var(--space-2)">
+    <div class="main-layout">
-      <Panel title="Stats" :status="status">
+      <!-- Left column: Pipeline control (full height) -->
-        <div class="stats" v-if="stats">
+      <div class="pipeline-col" :style="{ width: pipelineWidth + 'px' }">
+        <PipelineGraphPanel :source="source" :status="status" />
+      </div>
+      <ResizeHandle direction="horizontal" @resize="onPipelineResize" />
+
+      <!-- Right area: interactive panels -->
+      <div class="content-col">
+        <!-- Row 1: Frame viewer + Funnel -->
+        <div class="viewer-row" :style="{ height: viewerHeight + 'px' }">
+          <FramePanel :source="source" :status="status" />
+          <FunnelPanel :source="source" :status="status" />
+        </div>
+        <ResizeHandle direction="vertical" @resize="onViewerResize" />
+
+        <!-- Row 2: Detections + Stats side by side -->
+        <div class="detections-stats-row">
+          <div class="detections-col" :style="{ flex: detectionsFlex }">
+            <Panel title="Detections" :status="status">
+              <div class="detections-stack">
+                <div class="timeline-section" :style="{ flex: timelineFlex }">
+                  <TimelinePanel :source="source" :status="status" :embedded="true" />
+                </div>
+                <ResizeHandle direction="vertical" @resize="onTimelineResize" />
+                <div class="table-section" :style="{ flex: tableFlex }">
+                  <BrandTablePanel :source="source" :status="status" :embedded="true" />
+                </div>
+              </div>
+            </Panel>
+          </div>
+          <ResizeHandle direction="horizontal" @resize="onDetectionsResize" />
+          <div class="stats-col">
+            <Panel title="Pipeline" :status="status">
+              <div class="pipeline-stats">
                 <div class="stat" v-for="s in [
-            { label: 'Frames', value: stats.frames_extracted },
+                  { label: 'Frames', value: stats?.frames_extracted ?? '—' },
-            { label: 'After filter', value: stats.frames_after_scene_filter },
+                  { label: 'After filter', value: stats?.frames_after_scene_filter ?? '—' },
-            { label: 'Regions', value: stats.regions_detected },
+                  { label: 'Regions', value: stats?.regions_detected ?? '—' },
-            { label: 'OCR resolved', value: stats.regions_resolved_by_ocr },
+                  { label: 'OCR resolved', value: stats?.regions_resolved_by_ocr ?? '—' },
-            { label: 'Cloud calls', value: stats.cloud_llm_calls },
+                  { label: 'VLM escalated', value: stats?.regions_escalated_to_local_vlm ?? '—' },
-            { label: 'Cost', value: `$${stats.estimated_cloud_cost_usd.toFixed(4)}` },
+                  { label: 'Cloud escalated', value: stats?.regions_escalated_to_cloud_llm ?? '—' },
                 ]" :key="s.label">
                   <span class="label">{{ s.label }}</span>
                   <span class="value">{{ s.value }}</span>
                 </div>
               </div>
-        <div v-else class="empty">Waiting for stats...</div>
             </Panel>
-
-      <FunnelPanel :source="source" :status="status" />
-
-      <FramePanel :source="source" :status="status" />
-
-      <PipelineGraphPanel :source="source" :status="status" />
-
-      <BrandTablePanel :source="source" :status="status" />
-
-      <TimelinePanel :source="source" :status="status" />
-
             <CostStatsPanel :source="source" :status="status" />
+          </div>
+        </div>
+      </div>
+    </div>
 
+    <!-- Bottom: Log (full width) -->
+    <div class="log-row">
       <LogPanel :source="source" :status="status" />
-    </LayoutGrid>
+    </div>
   </div>
 </template>
 
@@ -92,10 +155,11 @@ body {
 
 .app {
   height: 100vh;
-  display: flex;
+  display: grid;
-  flex-direction: column;
+  grid-template-rows: auto 1fr auto;
   padding: var(--space-4);
   gap: var(--space-2);
+  overflow: hidden;
 }
 
 header {
@@ -120,21 +184,110 @@ header h1 { font-size: var(--font-size-lg); font-weight: 600; }
 .status-badge.live { background: var(--status-live); color: #000; }
 .status-badge.error { background: var(--status-error); color: #000; }
 
+.run-info {
+  color: var(--text-secondary);
+  font-size: var(--font-size-sm);
+}
+
 .job-id { color: var(--text-dim); font-size: var(--font-size-sm); margin-left: auto; }
 
-.stats {
+/* Main layout: pipeline left, content right — both same height */
-  display: grid;
+.main-layout {
-  grid-template-columns: repeat(3, 1fr);
+  display: flex;
   gap: var(--space-2);
+  min-height: 0;
+  overflow: hidden;
+}
+
+.pipeline-col {
+  flex-shrink: 0;
+  display: flex;
+  overflow: hidden;
+}
+
+.pipeline-col > * { flex: 1; }
+
+.content-col {
+  flex: 1;
+  display: flex;
+  flex-direction: column;
+  gap: var(--space-2);
+  min-height: 0;
+  overflow: hidden;
+}
+
+/* Content rows */
+.viewer-row {
+  display: flex;
+  gap: var(--space-2);
+  flex-shrink: 0;
+}
+.viewer-row > * { flex: 1; overflow: hidden; }
+
+/* Detections (75%) + Stats (25%) side by side, bottom-aligned with pipeline */
+.detections-stats-row {
+  display: flex;
+  gap: var(--space-2);
+  flex: 1;
+  min-height: 0;
+}
+
+.detections-col {
+  flex: 3;
+  min-width: 0;
+  display: flex;
+}
+.detections-col > * { flex: 1; display: flex; flex-direction: column; }
+
+.stats-col {
+  flex: 1;
+  display: flex;
+  flex-direction: column;
+  gap: var(--space-2);
+  min-width: 200px;
+}
+.stats-col > * { flex: 1; }
+
+.detections-stack {
+  display: flex;
+  flex-direction: column;
+  height: 100%;
+  overflow: hidden;
+}
+
+.timeline-section {
+  min-height: 60px;
+  overflow: auto;
+}
+
+.table-section {
+  min-height: 60px;
+  overflow-y: auto;
+  overflow-x: hidden;
+}
+
+/* Pipeline stats list */
+.pipeline-stats {
+  display: flex;
+  flex-direction: column;
+  gap: var(--space-1);
 }
 
 .stat {
-  background: var(--surface-2);
+  display: flex;
-  border-radius: var(--panel-radius);
+  justify-content: space-between;
-  padding: var(--space-3);
+  padding: var(--space-1) 0;
+}
+
+.stat .label { color: var(--text-dim); font-size: var(--font-size-sm); }
+.stat .value { font-weight: 600; }
+
+/* Log: full width bottom, fixed height */
+.log-row {
+  flex-shrink: 0;
+  height: 160px;
+  overflow: hidden;
 }
-.stat .label { display: block; color: var(--text-dim); font-size: var(--font-size-sm); margin-bottom: var(--space-1); }
-.stat .value { font-size: 20px; font-weight: 600; }
 
 .empty { color: var(--text-dim); padding: var(--space-6); text-align: center; }
 </style>

ui/detection-app/src/panels/BrandTablePanel.vue

@@ -8,6 +8,7 @@ import type { DataSource } from 'mpr-ui-framework'
 const props = defineProps<{
   source: DataSource
   status?: 'idle' | 'live' | 'processing' | 'error'
+  embedded?: boolean
 }>()
 
 const columns: TableColumn[] = [
@@ -45,7 +46,7 @@ function onSort(key: string) {
 </script>
 
 <template>
-  <Panel title="Detections" :status="status">
+  <Panel v-if="!embedded" title="Detections" :status="status">
     <TableRenderer
       :columns="columns"
       :rows="rows"
@@ -54,4 +55,12 @@ function onSort(key: string) {
       @sort="onSort"
     />
   </Panel>
+  <TableRenderer
+    v-else
+    :columns="columns"
+    :rows="rows"
+    :sort-key="sortKey"
+    :sort-dir="sortDir"
+    @sort="onSort"
+  />
 </template>

ui/detection-app/src/panels/CostStatsPanel.vue

@@ -7,6 +7,7 @@ import type { StatsUpdate, Detection } from '../types/sse-contract'
 const props = defineProps<{
   source: DataSource
   status?: 'idle' | 'live' | 'processing' | 'error'
+  embedded?: boolean
 }>()
 
 const stats = ref<StatsUpdate | null>(null)
@@ -71,7 +72,7 @@ const metrics = computed<Metric[]>(() => {
 </script>
 
 <template>
-  <Panel title="Cost & Stats" :status="status">
+  <Panel v-if="!embedded" title="Cost & Stats" :status="status">
     <div class="cost-stats" v-if="stats">
       <div class="metric" v-for="m in metrics" :key="m.label">
         <span class="label">{{ m.label }}</span>
@@ -81,38 +82,59 @@ const metrics = computed<Metric[]>(() => {
     </div>
     <div v-else class="empty">Waiting for stats...</div>
   </Panel>
+  <div v-else class="cost-stats" v-show="stats">
+    <div class="metric" v-for="m in metrics" :key="m.label">
+      <span class="label">{{ m.label }}</span>
+      <span class="value" :style="m.color ? { color: m.color } : {}">{{ m.value }}</span>
+      <span class="sub" v-if="m.sub">{{ m.sub }}</span>
+    </div>
+  </div>
 </template>
 
 <style scoped>
 .cost-stats {
   display: grid;
   grid-template-columns: 1fr 1fr;
-  gap: var(--space-3);
+  gap: var(--space-2);
-  padding: var(--space-3);
+  padding: var(--space-2);
+  overflow: hidden;
+  box-sizing: border-box;
+  width: 100%;
 }
 
 .metric {
   background: var(--surface-2);
   border-radius: var(--panel-radius);
-  padding: var(--space-3);
+  padding: var(--space-2);
   display: flex;
   flex-direction: column;
-  gap: var(--space-1);
+  gap: 2px;
+  overflow: hidden;
+  min-width: 0;
 }
 
 .label {
   font-size: var(--font-size-sm);
   color: var(--text-dim);
+  white-space: nowrap;
+  overflow: hidden;
+  text-overflow: ellipsis;
 }
 
 .value {
-  font-size: 22px;
+  font-size: 18px;
   font-weight: 600;
+  white-space: nowrap;
+  overflow: hidden;
+  text-overflow: ellipsis;
 }
 
 .sub {
-  font-size: 11px;
+  font-size: 10px;
   color: var(--text-dim);
+  white-space: nowrap;
+  overflow: hidden;
+  text-overflow: ellipsis;
 }
 
 .empty {

ui/detection-app/src/panels/TimelinePanel.vue

@@ -7,6 +7,7 @@ import type { Detection } from '../types/sse-contract'
 const props = defineProps<{
   source: DataSource
   status?: 'idle' | 'live' | 'processing' | 'error'
+  embedded?: boolean
 }>()
 
 interface TimelineEntry {
@@ -45,6 +46,20 @@ const maxTime = computed(() => {
   return Math.max(...entries.value.map((e) => e.timestamp + e.duration)) * 1.1
 })
 
+// Ruler ticks — nice intervals based on duration
+const rulerTicks = computed(() => {
+  const max = maxTime.value
+  const intervals = [1, 2, 5, 10, 15, 30, 60, 120, 300]
+  const targetTicks = 8
+  const rawStep = max / targetTicks
+  const step = intervals.find((i) => i >= rawStep) || Math.ceil(rawStep / 60) * 60
+  const ticks: number[] = []
+  for (let t = 0; t <= max; t += step) {
+    ticks.push(t)
+  }
+  return ticks
+})
+
 const sourceColor: Record<string, string> = {
   ocr: '#3ecf8e',
   local_vlm: '#f5a623',
@@ -67,8 +82,19 @@ function barStyle(entry: TimelineEntry) {
 </script>
 
 <template>
-  <Panel title="Detection Timeline" :status="status">
+  <Panel v-if="!embedded" title="Detection Timeline" :status="status">
     <div class="timeline" v-if="brands.length > 0">
+      <div class="ruler">
+        <span class="ruler-label" />
+        <div class="ruler-track">
+          <span
+            v-for="t in rulerTicks" :key="t"
+            class="ruler-tick"
+            :style="{ left: (t / maxTime) * 100 + '%' }"
+          >{{ t }}s</span>
+        </div>
+      </div>
+      <div class="rows">
         <div class="row" v-for="brand in brands" :key="brand">
           <span class="brand-label">{{ brand }}</span>
           <div class="track">
@@ -81,10 +107,6 @@ function barStyle(entry: TimelineEntry) {
             />
           </div>
         </div>
-      <div class="time-axis">
-        <span>0s</span>
-        <span>{{ (maxTime / 2).toFixed(0) }}s</span>
-        <span>{{ maxTime.toFixed(0) }}s</span>
       </div>
       <div class="legend">
         <span v-for="(color, source) in sourceColor" :key="source" class="legend-item">
@@ -95,6 +117,32 @@ function barStyle(entry: TimelineEntry) {
     </div>
     <div v-else class="empty">Waiting for detections...</div>
   </Panel>
+  <div v-else class="timeline" v-show="brands.length > 0">
+    <div class="ruler">
+      <span class="ruler-label" />
+      <div class="ruler-track">
+        <span
+          v-for="t in rulerTicks" :key="t"
+          class="ruler-tick"
+          :style="{ left: (t / maxTime) * 100 + '%' }"
+        >{{ t }}s</span>
+      </div>
+    </div>
+    <div class="rows">
+      <div class="row" v-for="brand in brands" :key="brand">
+        <span class="brand-label">{{ brand }}</span>
+        <div class="track">
+          <div
+            v-for="(entry, i) in entries.filter((e) => e.brand === brand)"
+            :key="i"
+            class="bar"
+            :style="barStyle(entry)"
+            :title="`${entry.brand} — ${entry.source} (${(entry.confidence * 100).toFixed(0)}%) @ ${entry.timestamp.toFixed(1)}s`"
+          />
+        </div>
+      </div>
+    </div>
+  </div>
 </template>
 
 <style scoped>
@@ -102,9 +150,54 @@ function barStyle(entry: TimelineEntry) {
   padding: var(--space-2);
   display: flex;
   flex-direction: column;
-  gap: var(--space-1);
   height: 100%;
-  overflow-y: auto;
+  overflow: auto;
+}
+
+.ruler {
+  display: flex;
+  align-items: flex-end;
+  gap: var(--space-2);
+  flex-shrink: 0;
+  height: 20px;
+  margin-bottom: var(--space-1);
+}
+
+.ruler-label {
+  width: 100px;
+  flex-shrink: 0;
+}
+
+.ruler-track {
+  flex: 1;
+  position: relative;
+  height: 100%;
+  border-bottom: 1px solid var(--surface-3);
+}
+
+.ruler-tick {
+  position: absolute;
+  bottom: 0;
+  transform: translateX(-50%);
+  font-size: 9px;
+  color: var(--text-dim);
+  padding-bottom: 2px;
+}
+
+.ruler-tick::after {
+  content: '';
+  position: absolute;
+  bottom: -1px;
+  left: 50%;
+  width: 1px;
+  height: 6px;
+  background: var(--surface-3);
+}
+
+.rows {
+  display: flex;
+  flex-direction: column;
+  gap: var(--space-1);
 }
 
 .row {
@@ -141,15 +234,6 @@ function barStyle(entry: TimelineEntry) {
   min-width: 4px;
 }
 
-.time-axis {
-  display: flex;
-  justify-content: space-between;
-  padding-left: 108px;
-  font-size: 10px;
-  color: var(--text-dim);
-  margin-top: var(--space-1);
-}
-
 .legend {
   display: flex;
   gap: var(--space-3);

ui/detection-app/src/types/sse-contract.ts

@@ -95,3 +95,44 @@ export interface JobComplete {
   job_id: string;
   report: DetectionReportSummary | null;
 }
+
+// --- Run context (injected into all SSE events) ---
+
+export interface RunContext {
+  run_id: string;
+  parent_job_id: string;
+  run_type: 'initial' | 'replay' | 'retry';
+}
+
+// --- Checkpoint API types ---
+
+export interface CheckpointInfo {
+  stage: string;
+}
+
+export interface ReplayRequest {
+  job_id: string;
+  start_stage: string;
+  config_overrides?: Record<string, unknown>;
+}
+
+export interface ReplayResponse {
+  status: string;
+  job_id: string;
+  start_stage: string;
+  detections: number;
+  brands_found: number;
+}
+
+export interface RetryRequest {
+  job_id: string;
+  config_overrides?: Record<string, unknown>;
+  start_stage?: string;
+  schedule_seconds?: number;
+}
+
+export interface RetryResponse {
+  status: string;
+  task_id: string;
+  job_id: string;
+}

ui/framework/src/components/Panel.vue

@@ -64,8 +64,9 @@ defineProps<{
 
 .panel-body {
   flex: 1;
-  overflow: auto;
+  overflow: hidden;
   padding: var(--space-2);
+  min-height: 0;
 }
 
 .panel-overlay {

ui/framework/src/components/ResizeHandle.vue

@@ -0,0 +1,70 @@
+<script setup lang="ts">
+import { ref } from 'vue'
+
+const props = defineProps<{
+  direction: 'horizontal' | 'vertical'
+}>()
+
+const emit = defineEmits<{
+  resize: [delta: number]
+}>()
+
+const dragging = ref(false)
+let startPos = 0
+
+function onPointerDown(e: PointerEvent) {
+  dragging.value = true
+  startPos = props.direction === 'horizontal' ? e.clientX : e.clientY
+  const el = e.target as HTMLElement
+  el.setPointerCapture(e.pointerId)
+}
+
+function onPointerMove(e: PointerEvent) {
+  if (!dragging.value) return
+  const currentPos = props.direction === 'horizontal' ? e.clientX : e.clientY
+  const delta = currentPos - startPos
+  startPos = currentPos
+  emit('resize', delta)
+}
+
+function onPointerUp() {
+  dragging.value = false
+}
+</script>
+
+<template>
+  <div
+    class="resize-handle"
+    :class="[direction, { dragging }]"
+    @pointerdown="onPointerDown"
+    @pointermove="onPointerMove"
+    @pointerup="onPointerUp"
+  />
+</template>
+
+<style scoped>
+.resize-handle {
+  flex-shrink: 0;
+  background: transparent;
+  transition: background 0.15s;
+  touch-action: none;
+  z-index: 10;
+}
+
+.resize-handle:hover,
+.resize-handle.dragging {
+  background: var(--text-dim);
+}
+
+.resize-handle.horizontal {
+  width: 4px;
+  cursor: col-resize;
+  margin: 0 -2px;
+}
+
+.resize-handle.vertical {
+  height: 4px;
+  cursor: row-resize;
+  margin: -2px 0;
+}
+</style>

ui/framework/src/index.ts

@@ -7,6 +7,7 @@ export { useDataSource } from './composables/useDataSource'
 // Components
 export { default as Panel } from './components/Panel.vue'
 export { default as LayoutGrid } from './components/LayoutGrid.vue'
+export { default as ResizeHandle } from './components/ResizeHandle.vue'
 
 // Renderers
 export { default as LogRenderer } from './renderers/LogRenderer.vue'

ui/framework/src/renderers/TableRenderer.vue

@@ -76,6 +76,7 @@ const sorted = computed(() => {
 table {
   width: 100%;
   border-collapse: collapse;
+  table-layout: fixed;
 }
 
 th {
@@ -89,7 +90,6 @@ th {
   border-bottom: var(--panel-border);
   cursor: pointer;
   user-select: none;
-  white-space: nowrap;
 }
 
 th:hover {
@@ -104,7 +104,10 @@ th:hover {
 td {
   padding: var(--space-1) var(--space-3);
   border-bottom: 1px solid var(--surface-3);
-  white-space: nowrap;
+  white-space: normal;
+  word-break: break-word;
+  overflow: hidden;
+  text-overflow: ellipsis;
 }
 
 tr:hover td {

ui/framework/src/renderers/TimeSeriesRenderer.vue

@@ -22,6 +22,7 @@ const props = withDefaults(defineProps<{
 })
 
 const container = ref<HTMLElement | null>(null)
+const zoomed = ref(false)
 let chart: uPlot | null = null
 
 function buildOpts(): uPlot.Options {
@@ -40,25 +41,66 @@ function buildOpts(): uPlot.Options {
     height: container.value?.clientHeight ?? 200,
     series: seriesOpts,
     axes: [
-      { stroke: '#555568', grid: { stroke: '#2e2e3822' } },
+      {
-      { stroke: '#555568', grid: { stroke: '#2e2e3822' } },
+        stroke: '#555568',
+        grid: { stroke: '#2e2e3822' },
+        size: 40,
+        font: '10px monospace',
+        ticks: { size: 3 },
+      },
+      {
+        stroke: '#555568',
+        grid: { stroke: '#2e2e3822' },
+        size: 35,
+        font: '10px monospace',
+        ticks: { size: 3 },
+      },
     ],
     cursor: { show: true },
-    legend: { show: true },
+    legend: { show: true, live: false },
+    padding: [8, 8, 0, 0],
+    hooks: {
+      setScale: [(_self: uPlot, scaleKey: string) => {
+        if (scaleKey === 'x') zoomed.value = true
+      }],
+    },
   }
 }
 
+function resetZoom() {
+  if (!chart) return
+  const data = chart.data
+  if (data && data[0] && data[0].length > 0) {
+    const min = data[0][0]
+    const max = data[0][data[0].length - 1]
+    chart.setScale('x', { min, max })
+  }
+  zoomed.value = false
+}
+
+function getLegendHeight(): number {
+  if (!container.value) return 0
+  const legend = container.value.querySelector('.u-legend') as HTMLElement | null
+  return legend ? legend.offsetHeight : 0
+}
+
 function createChart() {
   if (!container.value) return
   if (chart) chart.destroy()
   chart = new uPlot(buildOpts(), props.data, container.value)
+  // Refit after legend renders
+  nextTick(() => resize())
 }
 
 function resize() {
   if (!chart || !container.value) return
+  const legendH = getLegendHeight()
+  const availableH = container.value.clientHeight
+  // uPlot height = canvas height (chart sets total = canvas + legend)
+  const chartH = Math.max(60, availableH - legendH)
   chart.setSize({
     width: container.value.clientWidth,
-    height: container.value.clientHeight,
+    height: chartH,
   })
 }
 
@@ -83,19 +125,74 @@ onMounted(() => {
 </script>
 
 <template>
+  <div class="timeseries-wrapper">
+    <button v-if="zoomed" class="reset-zoom" @click="resetZoom" title="Reset zoom">⟲</button>
     <div ref="container" class="timeseries-renderer" />
+  </div>
 </template>
 
 <style scoped>
+.timeseries-wrapper {
+  width: 100%;
+  height: 100%;
+  position: relative;
+}
+
+.reset-zoom {
+  position: absolute;
+  top: 4px;
+  right: 4px;
+  z-index: 20;
+  background: var(--surface-2);
+  border: 1px solid var(--surface-3);
+  border-radius: 4px;
+  color: var(--text-secondary);
+  font-size: 14px;
+  width: 24px;
+  height: 24px;
+  cursor: pointer;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  opacity: 0.7;
+  transition: opacity 0.15s;
+}
+
+.reset-zoom:hover {
+  opacity: 1;
+  color: var(--text-primary);
+}
+
 .timeseries-renderer {
   width: 100%;
   height: 100%;
-  min-height: 150px;
+  display: flex;
+  flex-direction: column;
+  overflow: hidden;
+}
+
+/* uPlot creates a .u-wrap for canvas + a .u-legend below it */
+.timeseries-renderer :deep(.u-wrap) {
+  flex: 1;
+  min-height: 0;
+}
+
+.timeseries-renderer :deep(.u-legend) {
+  flex-shrink: 0;
 }
 
 .timeseries-renderer :deep(.u-legend) {
   font-family: var(--font-mono);
-  font-size: var(--font-size-sm);
+  font-size: 10px;
   color: var(--text-secondary);
+  padding: 2px 0;
+  display: flex;
+  flex-wrap: wrap;
+  gap: 0 8px;
+}
+
+.timeseries-renderer :deep(.u-legend .u-series) {
+  display: inline-flex;
+  padding: 0;
 }
 </style>