Harden detection result review

docs/AI_PIPELINES.md

@@ -352,6 +352,16 @@ remains available as a higher-precision legacy `0.15` choice. The older
 default-promotion blocker. Every production-like run still requires persisted
 QA/QC against suitable reference data.
 
+The persisted seven-AOI evidence for this profile contains 5,568 false
+positives among 13,613 candidate detections. The read-only audit command in
+`scripts/README.md` reports the largest review volumes in Turnhout, Herentals
+and Geel, a median false-positive geometry area of about 184.5 m2, and 25.8%
+tiny/small geometry below 100 m2. Current evidence does not include
+per-detection confidence, so model-review reports must retain confidence
+coverage as zero rather than treating threshold `0.15` as an observed score.
+Combined false-positive GeoJSON is evidence for operator review only; a feature
+must be visually confirmed before it is used as a hard-negative label.
+
 To update a Tower/Unraid `.env` from a promoted report, use the guarded
 activation helper. It validates the exact report candidate key, verifies that
 the candidate has `promotion_status=promote_candidate`, resolves the local model

docs/CODEX_EXECUTION_LOG.md

@@ -7138,3 +7138,41 @@ Open:
 ## Next recommended pass
 
 - After redeploy, verify the active model SHA, local model-load preflight, live PostGIS migration smoke and browser profile selection. Then review false-positive evidence and the remaining 5,838 persistent misses before any further training.
+
+# Sprint 175 - Detection result scale and false-positive evidence review
+
+## UI hardening
+
+- Confirmed that Detection Lab rendered every persisted detection row at once; Westerlo alone produced 1,172 body rows in the browser.
+- Added local 25/50/100-row pagination with a default of 50 rows, bounded page controls and automatic page-one reset after run/filter/result changes.
+- Kept the full persisted collection unchanged for the existing MapLibre GeoJSON overlay and detection QA/QC. No endpoint, response envelope or persistence contract changed.
+
+## Persisted false-positive evidence
+
+- Added `scripts/audit_detection_false_positive_evidence.py` as a read-only evidence consumer.
+- The audit validates FeatureCollection/polygon geometry, compares declared portfolio role counts with actual evidence, computes WGS84 geodesic areas, preserves original feature provenance and emits combined `false_positives.geojson`.
+- Source tile summaries are qualified by AOI because `tile_index` is local to each raster manifest.
+- The active fixed-threshold seven-AOI portfolio produced:
+  - 5,568 false positives among 13,613 candidate detections (`0.4090` false-positive share);
+  - median false-positive geometry area `184.5 m2`, p90 `607.7 m2`;
+  - 52 below 25 m2, 1,382 between 25-100 m2, 3,412 between 100-500 m2 and 722 at or above 500 m2;
+  - largest AOI review volumes: Turnhout `1,102`, Herentals `917`, Geel `913`;
+  - largest AOI-qualified tile hotspot: `turnhout:0` with 236 false positives.
+- Existing persisted QA evidence carries run threshold and tile index but no per-detection confidence. The audit reports confidence coverage `0/5,568` and does not invent confidence statistics.
+- Added focused regression coverage, readiness compilation and all-in-one image inclusion. No training, inference, provider fetch, model download or activation occurred.
+
+## Local validation
+
+- `python -m compileall backend/app`: passed.
+- `python -m pytest`: 475 passed.
+- `python -m ruff check` for the new audit/test modules: passed.
+- `npm run typecheck`: passed.
+- `npm run build`: passed; app bundle `216.83 kB`, MapLibre bundle `801.82 kB` before gzip.
+- `bash scripts/run_readiness_check.sh`: passed with 475 tests and all release-critical syntax gates.
+- `python -m alembic heads`: one head, `202606120900`.
+- `python -m alembic upgrade head --sql`: complete migration chain rendered successfully.
+- Local `docker compose config` could not run because Docker CLI is not installed on the Windows host; live image/PostGIS validation is delegated to the Docker-enabled Tower deployment.
+
+## Next recommended pass
+
+- Redeploy and verify bounded table rendering against the live 1,172-detection Westerlo run. Then visually classify a stratified sample from Turnhout, Herentals and Geel before deciding whether any confirmed false positives belong in a new hard-negative corpus.

docs/TODO.md

@@ -138,7 +138,8 @@ This file now starts with the current implementation status. Older preparation/b
 - [x] Review per-AOI false-negative evidence, expand positive sample/label coverage and verify the resulting candidate improves false-negative rate in every validated AOI.
 - [x] Complete rebuild/restart and browser/runtime smoke for the guarded promoted V1 building detector activation.
 - [x] Expand focused small-building training evidence after reviewing persistent false negatives, train one inactive candidate and pass it through positive, pure-empty and fixed-reference promotion evidence before guarded activation.
-- [ ] Review the remaining 5,838 persistent false negatives and the increased false-positive load before any further model training; do not start another blind run.
+- [x] Audit the promoted model's increased false-positive load from persisted seven-AOI QA evidence, including geodetic area buckets, AOI-qualified tile hotspots and combined review GeoJSON.
+- [ ] Visually classify representative false-positive evidence from Turnhout, Herentals and Geel and review the remaining 5,838 persistent false negatives before any further model training; do not start another blind run.
 
 ## Sprint 8 status