feat(agent): crop proposers — booru_yolo anatomy + COCO person + comic panels (#1202)

Better region PROPOSERS feeding the existing crop→SigLIP→max-over-bag heads (no
change to the learned-tagging approach; no per-tag cost — propose once, embed
each region, all heads in one matmul).

- detectors.py: lazy ultralytics YOLO wrapper, each proposer independently
  optional + guarded (a bad weight spec / inference error self-disables that one,
  logged, never breaks the worker). Weights resolve from an ultralytics name |
  http(s) URL | "hf_repo::file", cached under HF_HOME. NMS merge so a figure two
  detectors both find collapses to one crop.
- worker: figure boxes = imgutils detect_person ∪ general COCO person (merged)
  → CCIP + concept (anime + Western/realistic coverage); booru_yolo anatomy
  components (head/cat-head/anatomy/…) → concept crops; comic panels → kind=
  'panel' concept crops. Capped per frame (MAX_COMPONENTS/MAX_PANELS).
- config + compose: PERSON_WEIGHTS (default yolo11n.pt, works OOB),
  ANATOMY_WEIGHTS + PANEL_WEIGHTS (operator sets booru_yolo URL + mosesb panel
  hf::file; empty = off). ultralytics added to requirements.
- backend: image_region 'kind' doc notes 'panel'; no migration (free String,
  and the bag scorer keys on a non-null siglip_embedding, not the kind, so any
  SigLIP region joins the bag automatically).

Agent is outside CI — py-compiled here; operator tests on the GPU and checks
Western-vs-anime crop quality via /api/ccip observability.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa

agent/fc_agent/worker.py

@@ -76,6 +76,9 @@ class Worker:
         # needs it, from the model the server announces — one shared instance.
         self._embedder = None
         self._embedder_lock = threading.Lock()
+        # Region proposers (extra YOLO detectors) — lazily built once, shared.
+        self._proposers = None
+        self._proposers_lock = threading.Lock()
 
     # --- control -----------------------------------------------------------
     def start(self):
@@ -177,6 +180,15 @@ class Worker:
                 self._embedder = CropEmbedder(model_name, self.cfg.embed_dtype)
         return self._embedder
 
+    def _ensure_proposers(self):
+        if self._proposers is not None:
+            return self._proposers
+        with self._proposers_lock:
+            if self._proposers is None:
+                from .detectors import Proposers
+                self._proposers = Proposers(self.cfg)
+        return self._proposers
+
     def _process(self, job: dict) -> bool:
         """Process one job. Returns True when handled (completed, or hard-failed
         because the job itself is bad) and False on a TRANSPORT error (curator
@@ -227,43 +239,68 @@ class Worker:
             want_ccip = task in ("ccip", "both")
             want_siglip = task in ("ccip", "siglip", "both")
             replace_kinds = (
-                ["concept"] if task == "siglip" else ["figure", "face", "concept"]
+                ["concept", "panel"] if task == "siglip"
+                else ["figure", "face", "concept", "panel"]
             )
             embedder = self._ensure_embedder(model_name) if want_siglip else None
+            proposers = self._ensure_proposers()
 
             regions = []
             ccip_ev = self.cfg.ccip_model or "ccip-default"
             dv = f"person-{self.cfg.detector_level}"
+
+            def _concept(frame, bbox, t, score, detver, kind="concept"):
+                """A SigLIP region for one crop (None if below the size floor)."""
+                crop = crop_region(frame, bbox)
+                if crop is None:
+                    return None
+                return {
+                    "kind": kind, "bbox": list(bbox), "frame_time": t,
+                    "score": score, "siglip_embedding": embedder.embed(crop),
+                    "embedding_version": embed_version, "detector_version": detver,
+                }
+
             for t, frame in frames:
-                figs = models.detect_figures(frame, self.cfg.detector_level)
+                # FIGURE boxes: imgutils detect_person ∪ general COCO person,
+                # NMS-merged → CCIP identity (+ a concept crop). Covers anime +
+                # Western/realistic figures.
+                base = models.detect_figures(frame, self.cfg.detector_level)
+                figs = proposers.figures(frame, base)
                 if not figs:
-                    figs = [((0.0, 0.0, 1.0, 1.0), None)]   # whole-frame fallback
-                for bbox, score in figs:
+                    figs = [((0.0, 0.0, 1.0, 1.0), 1.0, "whole")]  # whole-frame fallback
+                for bbox, score, _label in figs:
                     crop = crop_region(frame, bbox)
                     if crop is None:
                         continue
                     if want_ccip:
                         regions.append({
-                            "kind": "figure",
-                            "bbox": list(bbox),
-                            "frame_time": t,
+                            "kind": "figure", "bbox": list(bbox), "frame_time": t,
                             "score": score,
                             "ccip_embedding": models.ccip_vector(
                                 crop, self.cfg.ccip_model or None
                             ),
-                            "embedding_version": ccip_ev,
-                            "detector_version": dv,
+                            "embedding_version": ccip_ev, "detector_version": dv,
                         })
                     if want_siglip:
                         regions.append({
-                            "kind": "concept",
-                            "bbox": list(bbox),
-                            "frame_time": t,
+                            "kind": "concept", "bbox": list(bbox), "frame_time": t,
                             "score": score,
                             "siglip_embedding": embedder.embed(crop),
-                            "embedding_version": embed_version,
-                            "detector_version": dv,
+                            "embedding_version": embed_version, "detector_version": dv,
                         })
+                if not want_siglip:
+                    continue
+                # ANATOMY components (booru_yolo: head/cat-head/anatomy/…) →
+                # concept crops only (not full characters, so no CCIP).
+                for bbox, score, label in proposers.components(frame):
+                    r = _concept(frame, bbox, t, score, f"booru:{label}")
+                    if r is not None:
+                        regions.append(r)
+                # PANEL crops (comic page → panels) → kind='panel' (still SigLIP).
+                for bbox, score, _label in proposers.panels(frame):
+                    r = _concept(frame, bbox, t, score, "panel", kind="panel")
+                    if r is not None:
+                        regions.append(r)
             self.client.submit(job["job_id"], regions, replace_kinds)
             self._bump(processed=1)
             return True