Promote focused small-building detector

docs/AI_PIPELINES.md

@@ -335,17 +335,22 @@ hard-negative gates, then run `sparse_building_context` as a separate review
 matrix. The first expanded local model improved dense AOI F1, but Kasterlee-bos
 false positives block default promotion.
 
-The expanded-AOI local model asset,
-`geointel-building-yolov8s-aoi1024expandedminpx4vis035e50-pt`, is the current
+The focused small-building local model asset,
+`geointel-building-yolov8s-smallbld-minpx3-img640-ft30-pt`, is the current
 recommended Detection Lab operator profile. Use tile size `512`, overlap `64`
-and confidence threshold `0.15`. Persisted QA/QC across seven positive AOIs
-measured mean precision `0.6471`, recall `0.4700` and F1 `0.5433`; the strict
-three-sample pure-empty background gate produced zero detections. The previous
-`geointel-building-yolov8s-aoi1024bg512r3e50-pt` model remains available as a
-legacy conservative `0.35` review profile. Sparse-context detections remain
-review-only evidence, not a default-promotion blocker. Persistent misses are
-concentrated in small buildings, so every production-like run still requires
-persisted QA/QC against suitable reference data.
+and confidence threshold `0.15`. Persisted QA/QC at match IoU `0.25` across
+seven positive AOIs measured mean precision `0.5898`, recall `0.5770` and F1
+`0.5825`; minimum per-AOI F1 was `0.5528`. The strict three-sample pure-empty
+background gate produced zero detections. Compared with the previous balanced
+profile, the same persisted reference populations contain 1,571 fewer false
+negatives, including 745 fewer misses in the 25-100 m2 bucket and 181 fewer
+below 25 m2. This recall gain increases the false-positive review load, so the
+previous `geointel-building-yolov8s-aoi1024expandedminpx4vis035e50-pt` profile
+remains available as a higher-precision legacy `0.15` choice. The older
+`geointel-building-yolov8s-aoi1024bg512r3e50-pt` remains the conservative
+`0.35` profile. Sparse-context detections remain review-only evidence, not a
+default-promotion blocker. Every production-like run still requires persisted
+QA/QC against suitable reference data.
 
 To update a Tower/Unraid `.env` from a promoted report, use the guarded
 activation helper. It validates the exact report candidate key, verifies that
@@ -355,8 +360,8 @@ when `--apply` is supplied:
 
 ```bash
 python scripts/activate_promoted_yolo_candidate.py \
-  --promotion-report artifacts/detection-model-promotion/split-aware/aoi1024expandedminpx4vis035e50-split/detection_model_promotion_report.json \
-  --candidate-key 'geointel-building-yolov8s-aoi1024expandedminpx4vis035e50-pt|512|64|0.15' \
+  --promotion-report storage/operator-data/model-review/small-building-candidate/promotion/detection_model_promotion_report.json \
+  --candidate-key 'geointel-building-yolov8s-smallbld-minpx3-img640-ft30-pt|512|64|0.15' \
   --models-dir /mnt/user/appdata/geointel/models \
   --env-file /mnt/user/appdata/geointel/.env \
   --json
@@ -440,6 +445,13 @@ and records the positive/negative tile counts. This gives the training smoke
 more image samples while preserving the same explicit operator-data and QA/QC
 validation boundary.
 
+Focused small-building experiments use Beerse, Rijkevorsel, Hoogstraten and
+Vorselaar as training AOIs, with Vosselaar and Grobbendonk retained as
+independent validation AOIs. The exporter accepts an explicit `--samples`
+subset and records `source_manifest_sample_count`, `selected_sample_slugs` and
+`excluded_sample_slugs` in its summary. Manifest-backed validation samples
+cannot silently enter training.
+
 For visual error inspection, export the persisted QA evidence from a calibration
 summary: