Jens/geointel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Gate YOLO tile labels by visible ratio
GeoIntel CI / docs-smoke (push) Has been cancelled
Details
GeoIntel CI / contract-smoke (push) Has been cancelled
Details

Browse Source
This commit is contained in:
Codex
2026-07-09 13:12:33 +02:00
parent c4a2e13d43
commit a20d9b70c7
8 changed files with 93 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/CODEX_EXECUTION_LOG.md
+24
View File
@@ -1,3 +1,27 @@
## Sprint 150 YOLO label visible-ratio gate (2026-07-09)
Changed:
- Added `--min-label-visible-ratio` / `OPERATOR_YOLO_MIN_LABEL_VISIBLE_RATIO` to `scripts/export_operator_yolo_tile_dataset.py`.
- The tile exporter now computes the visible share of each original building bbox inside a tile and can drop labels below the configured ratio.
- Default remains `0` for legacy behavior; use `0.25` for the next overlap-heavy operator dataset experiment.
- Tile dataset summaries include `min_label_visible_ratio`.
- `scripts/audit_operator_yolo_dataset_quality.py` now reports `min_label_visible_ratio` in JSON and Markdown.
- Updated operator script documentation.
Why:
- The current rejected AOI512 candidate still shows low precision/recall after max-det and duplicate suppression hardening.
- A likely label-quality issue is that overlapping tile export can create many small clipped edge labels for buildings mostly outside a tile.
- This pass improves the next training dataset gate without activating a model, faking detections, fetching providers or changing persistence.
Tested:
- Red step: `python -m pytest backend\tests\test_sprint130_operator_yolo_tile_dataset.py -q` failed because the exporter lacked `min_label_visible_ratio`, CLI help and visible-fragment filtering.
- `python -m pytest backend\tests\test_sprint130_operator_yolo_tile_dataset.py -q` (`6 passed`)
- Red step: `python -m pytest backend\tests\test_sprint146_operator_yolo_dataset_quality_audit.py -q` failed because the audit report did not expose `min_label_visible_ratio`.
- `python -m pytest backend\tests\test_sprint130_operator_yolo_tile_dataset.py backend\tests\test_sprint146_operator_yolo_dataset_quality_audit.py -q` (`7 passed`)
Next:
- Run full readiness, deploy Tower, export a new visible-ratio-gated operator tile dataset, audit it, then decide whether it is good enough for another CPU training candidate.
## Sprint 149 YOLO duplicate suppression evidence (2026-07-09)
Changed: