a

core/schema/serializers/pipeline.py

@@ -2,7 +2,7 @@
 Serializers for detection pipeline runtime models.
 
 Special handling:
-  - Frame.image (np.ndarray → S3, excluded from JSON)
+  - Frame.image (np.ndarray, ephemeral — only metadata serialized)
   - TextCandidate.frame (object ref → frame_sequence integer)
 Everything else uses dataclasses.asdict() via safe_construct.
 """
@@ -24,7 +24,7 @@ from ._common import safe_construct, serialize_dataclass, serialize_dataclass_li
 
 
 # ---------------------------------------------------------------------------
-# Frame — image goes to S3 separately
+# Frame — metadata only (image is ephemeral, re-extracted from chunks)
 # ---------------------------------------------------------------------------
 
 def serialize_frame_meta(frame: Frame) -> dict:
@@ -34,21 +34,9 @@ def serialize_frame_meta(frame: Frame) -> dict:
     return result
 
 
-def serialize_frames_with_upload(frames: list[Frame], job_id: str) -> tuple[list[dict], dict[int, str]]:
-    """Upload frame images to S3, return metadata + manifest."""
-    from core.detect.checkpoint.frames import save_frames
-
-    manifest = save_frames(job_id, frames)
-    meta = [serialize_frame_meta(f) for f in frames]
-    return meta, manifest
-
-
-def deserialize_frames_with_download(meta: list[dict], manifest: dict, job_id: str) -> list[Frame]:
-    """Load frames from S3 + metadata."""
-    from core.detect.checkpoint.frames import load_frames
-
-    int_manifest = {int(k): v for k, v in manifest.items()}
-    return load_frames(int_manifest, meta)
+def serialize_frames_meta(frames: list[Frame]) -> list[dict]:
+    """Serialize frame metadata for all frames."""
+    return [serialize_frame_meta(f) for f in frames]
 
 
 # ---------------------------------------------------------------------------