expand AI-assisted review to full building corpus

artifacts/evidence/accuracy/model-training/20260809-reviewedexp6-ai-assisted-review-ledger.json

@@ -1,7 +1,7 @@
 {
   "schema_version": 1,
-  "ledger_id": "reviewedexp6-ai-assisted-review-r1",
+  "ledger_id": "reviewedexp6-ai-assisted-review-r2-full-corpus",
-  "reviewed_at": "2026-08-09T14:00:00+02:00",
+  "reviewed_at": "2026-08-09T15:00:00+02:00",
   "reviewer": {
     "id": "openai-codex",
     "type": "ai_assistant",
@@ -10,16 +10,21 @@
   "status": "complete_ai_assisted_experimental_only",
   "explicit_limitations": [
     "This ledger is not human review and must never satisfy a human-review release gate.",
-    "Decisions cover the 64 rendered samples on the contact sheet, not every label instance in the corpus.",
+    "Decisions cover all 252 rendered tiles, but visual review still does not adjudicate every individual label as human ground truth.",
     "Visible roof geometry can be displaced from authoritative GRB ground-level footprints.",
     "Accepted samples are eligible only for experimental training and non-protected validation."
   ],
   "artifacts": {
-    "contact_sheet_path": "/app/storage/operator-data/model-review/reviewedexp6-corpus/visual-review/contact_sheet_001.png",
+    "review_summary_path": "/app/storage/operator-data/model-review/reviewedexp6-corpus/full-ai-review-r1/operator_yolo_label_qa_summary.json",
-    "contact_sheet_sha256": "411ee541dcd53b59abc1c12f733f164b28478a1398ab2d50be8c728b11c8fdc4",
+    "review_summary_sha256": "467e5c3c489c42e42872a63fbb0bba7c613824631598d86ea6324b22c007f75a",
-    "review_summary_path": "/app/storage/operator-data/model-review/reviewedexp6-corpus/visual-review/operator_yolo_label_qa_summary.json",
+    "contact_sheets": [
-    "rendered_tile_count": 64,
+      {"path":"/app/storage/operator-data/model-review/reviewedexp6-corpus/full-ai-review-r1/contact_sheet_001.png","sha256":"688b97a6e82e6b906b7966ed3e2b43685786e89d50549188bfdcc3837d042eb2","tile_count":64},
-    "rendered_label_count": 21611,
+      {"path":"/app/storage/operator-data/model-review/reviewedexp6-corpus/full-ai-review-r1/contact_sheet_002.png","sha256":"87f1945df3324dc100e08febd3e45f291a2d709ddebd9c627b082f30fd8f2d26","tile_count":64},
+      {"path":"/app/storage/operator-data/model-review/reviewedexp6-corpus/full-ai-review-r1/contact_sheet_003.png","sha256":"1873ab25b8ee843abe28d00329fe28998a7b39f9fdbcd8f8acfb58cc576cdf10","tile_count":64},
+      {"path":"/app/storage/operator-data/model-review/reviewedexp6-corpus/full-ai-review-r1/contact_sheet_004.png","sha256":"e64c252aac08792a76a62fe734728e7c81ba6a649ad2a1cd2051b4df1f95be98","tile_count":60}
+    ],
+    "rendered_tile_count": 252,
+    "rendered_label_count": 79192,
     "selected_sample_count": 28
   },
   "visual_criteria": [
@@ -30,34 +35,34 @@
     "dense roof-versus-ground-footprint disagreement is retained as an explicit limitation"
   ],
   "decisions": [
-    {"sample_slug":"arendonk_center","sampled_tiles":2,"sampled_labels":1163,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"arendonk_center","sampled_tiles":9,"sampled_labels":4172,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"balen","sampled_tiles":2,"sampled_labels":794,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"balen","sampled_tiles":9,"sampled_labels":2658,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"beerse_center","sampled_tiles":2,"sampled_labels":1158,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"beerse_center","sampled_tiles":9,"sampled_labels":4303,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"dessel_center","sampled_tiles":2,"sampled_labels":841,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"dessel_center","sampled_tiles":9,"sampled_labels":3405,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"dessel_heide","sampled_tiles":2,"sampled_labels":68,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
+    {"sample_slug":"dessel_heide","sampled_tiles":9,"sampled_labels":192,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
-    {"sample_slug":"geel","sampled_tiles":2,"sampled_labels":1244,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"geel","sampled_tiles":9,"sampled_labels":4615,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"geel_bel","sampled_tiles":2,"sampled_labels":62,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
+    {"sample_slug":"geel_bel","sampled_tiles":9,"sampled_labels":168,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
-    {"sample_slug":"grobbendonk_center","sampled_tiles":2,"sampled_labels":1008,"split":"val","category":"reference_aoi","decision":"non_protected_validation_only"},
+    {"sample_slug":"grobbendonk_center","sampled_tiles":9,"sampled_labels":2793,"split":"val","category":"reference_aoi","decision":"non_protected_validation_only"},
-    {"sample_slug":"herentals","sampled_tiles":2,"sampled_labels":1512,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"herentals","sampled_tiles":9,"sampled_labels":5087,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"herenthout_bos","sampled_tiles":2,"sampled_labels":199,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
+    {"sample_slug":"herenthout_bos","sampled_tiles":9,"sampled_labels":520,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
-    {"sample_slug":"hoogstraten_center","sampled_tiles":2,"sampled_labels":1063,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"hoogstraten_center","sampled_tiles":9,"sampled_labels":3761,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"hulshout_center","sampled_tiles":2,"sampled_labels":895,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"hulshout_center","sampled_tiles":9,"sampled_labels":3606,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"kasterlee_bos","sampled_tiles":2,"sampled_labels":72,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
+    {"sample_slug":"kasterlee_bos","sampled_tiles":9,"sampled_labels":137,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
-    {"sample_slug":"kasterlee_center","sampled_tiles":2,"sampled_labels":983,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"kasterlee_center","sampled_tiles":9,"sampled_labels":4022,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"laakdal_center","sampled_tiles":2,"sampled_labels":156,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible_with_dense_alignment_limitation"},
+    {"sample_slug":"laakdal_center","sampled_tiles":9,"sampled_labels":287,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible_with_dense_alignment_limitation"},
-    {"sample_slug":"lille_center","sampled_tiles":2,"sampled_labels":961,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"lille_center","sampled_tiles":9,"sampled_labels":3949,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"lommel_heide","sampled_tiles":6,"sampled_labels":0,"split":"train","category":"pure_empty_negative","decision":"experimental_training_eligible"},
+    {"sample_slug":"lommel_heide","sampled_tiles":9,"sampled_labels":0,"split":"train","category":"pure_empty_negative","decision":"experimental_training_eligible"},
-    {"sample_slug":"meerhout_bos","sampled_tiles":2,"sampled_labels":111,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
+    {"sample_slug":"meerhout_bos","sampled_tiles":9,"sampled_labels":206,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
-    {"sample_slug":"meerhout_center","sampled_tiles":2,"sampled_labels":1016,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"meerhout_center","sampled_tiles":9,"sampled_labels":4002,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"mol","sampled_tiles":2,"sampled_labels":951,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"mol","sampled_tiles":9,"sampled_labels":3666,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"nijlen_center","sampled_tiles":2,"sampled_labels":1309,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"nijlen_center","sampled_tiles":9,"sampled_labels":4900,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"olen_center","sampled_tiles":2,"sampled_labels":971,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"olen_center","sampled_tiles":9,"sampled_labels":3955,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"oud_turnhout_center","sampled_tiles":2,"sampled_labels":1535,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible_with_dense_alignment_limitation"},
+    {"sample_slug":"oud_turnhout_center","sampled_tiles":9,"sampled_labels":5492,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible_with_dense_alignment_limitation"},
-    {"sample_slug":"postel_bos","sampled_tiles":6,"sampled_labels":0,"split":"train","category":"pure_empty_negative","decision":"experimental_training_eligible"},
+    {"sample_slug":"postel_bos","sampled_tiles":9,"sampled_labels":0,"split":"train","category":"pure_empty_negative","decision":"experimental_training_eligible"},
-    {"sample_slug":"ravels_bos","sampled_tiles":2,"sampled_labels":22,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
+    {"sample_slug":"ravels_bos","sampled_tiles":9,"sampled_labels":55,"split":"train","category":"sparse_building_context","decision":"experimental_training_eligible"},
-    {"sample_slug":"rijkevorsel_center","sampled_tiles":2,"sampled_labels":1153,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"rijkevorsel_center","sampled_tiles":9,"sampled_labels":4231,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"vorselaar_center","sampled_tiles":2,"sampled_labels":930,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
+    {"sample_slug":"vorselaar_center","sampled_tiles":9,"sampled_labels":3858,"split":"train","category":"reference_aoi","decision":"experimental_training_eligible"},
-    {"sample_slug":"vosselaar_center","sampled_tiles":2,"sampled_labels":1434,"split":"val","category":"reference_aoi","decision":"non_protected_validation_only"}
+    {"sample_slug":"vosselaar_center","sampled_tiles":9,"sampled_labels":5152,"split":"val","category":"reference_aoi","decision":"non_protected_validation_only"}
   ],
   "accounting": {
     "accepted_experimental_training_samples": 26,
@@ -67,5 +72,13 @@
     "sparse_context_samples": 6,
     "human_review_complete": false,
     "production_release_eligible": false
+  },
+  "nested_box_audit": {
+    "exact_duplicate_pairs": 0,
+    "near_duplicate_pairs_iou_at_least_0_90": 0,
+    "possible_nested_pairs_containment_at_least_0_98": 43,
+    "flagged_tile_count": 36,
+    "share_of_rendered_labels": 0.000543,
+    "action": "requires_human_adjudication_before_release; no automatic rewrite or exclusion"
   }
 }

docs/CODEX_EXECUTION_LOG.md

@@ -12549,3 +12549,16 @@ Open:
   review and therefore does not accept this AI ledger as release evidence.
 - Did not add an override, fixture claim or synthetic reviewer identity. No
   model training, activation or production redeploy followed from this review.
+
+### Full-corpus extension
+
+- Added deterministic contact-sheet pagination and rendered all 252 retained
+  tiles across four pages. Accounting covers all 79,192 labels with zero
+  missing images/files, invalid rows or low-variance tiles.
+- Inspected every page at original resolution. Pure-empty tiles remain visibly
+  empty; sparse contexts remain distinct; dense tiles retain the documented
+  roof-versus-ground-footprint limitation.
+- A read-only same-class box probe found zero exact duplicates and zero pairs
+  at IoU >= 0.90. It flagged 43 possible containment pairs across 36 tiles for
+  human adjudication. No source label was modified or silently excluded.
+- Targeted renderer tests passed: 6 tests.

docs/TODO.md

@@ -1150,6 +1150,11 @@ This file now starts with the current implementation status. Older preparation/b
 - [x] Complete an explicitly AI-assisted per-AOI review of the 64-tile
   reviewedexp6 contact sheet, retaining reviewer type, artifact hash, split
   role, sample accounting and roof/ground-footprint limitations.
+- [x] Extend AI-assisted review to all 252 tiles and 79,192 labels through
+  deterministic four-page contact-sheet rendering with complete accounting.
+- [ ] Human-adjudicate the 43 possible nested same-class box pairs in 36 tiles;
+  do not rewrite or exclude them automatically because GRB objects can overlap
+  legitimately.
 - [ ] Convert the AI-assisted ledger into no stronger claim than experimental
   triage; a real human must independently review and sign the frozen artifacts
   before the governed training wrapper may unlock.

docs/reviews/2026-08-09-ai-assisted-checkpoint-review.md

@@ -103,3 +103,11 @@ It explicitly cannot satisfy the human-review release gate. The production
 training wrapper reproduced that boundary and no bypass was added. The exact
 ledger is retained in
 `artifacts/evidence/accuracy/model-training/20260809-reviewedexp6-ai-assisted-review-ledger.json`.
+
+The renderer was subsequently paginated and the review expanded from the
+64-tile sample to all 252 retained tiles. Four immutable pages now account for
+all 79,192 labels with zero missing images, missing label files, invalid rows
+or low-variance tiles. No exact or IoU>=0.90 duplicate box pair was found. A
+separate containment probe identified 43 potentially nested pairs across 36
+tiles (0.0543% relative to rendered labels); these remain human-adjudication
+candidates and were neither rewritten nor automatically excluded.

scripts/README.md

@@ -6,6 +6,11 @@ model and dataset hashes, standard Ultralytics detection metrics and a separate
 pure-background detection count. The output claim is validation ranking only;
 the script neither reads protected test data nor promotes a model.
 
+`render_operator_yolo_label_qa_contact_sheets.py` paginates complete visual
+reviews with `--tiles-per-sheet` (default `64`). This keeps large corpora
+inspectable while preserving deterministic tile selection, ordering, label
+accounting and stable `contact_sheet_001.png` naming for the first page.
+
 Setup-, import-, demo- en maintenance-scripts voor GeoIntel.
 
 ## WALOUS source provisioning

scripts/render_operator_yolo_label_qa_contact_sheets.py

@@ -19,7 +19,7 @@ from PIL import Image, ImageDraw, ImageFont
 
 JSON_NAME = "operator_yolo_label_qa_summary.json"
 MARKDOWN_NAME = "operator_yolo_label_qa_contact_sheet.md"
-CONTACT_SHEET_NAME = "contact_sheet_001.png"
+CONTACT_SHEET_NAME_TEMPLATE = "contact_sheet_{index:03d}.png"
 
 
 def parse_args() -> argparse.Namespace:
@@ -37,6 +37,12 @@ def parse_args() -> argparse.Namespace:
     )
     parser.add_argument("--columns", type=int, default=4, help="Contact-sheet columns")
     parser.add_argument("--thumb-size", type=int, default=256, help="Rendered tile thumbnail size in pixels")
+    parser.add_argument(
+        "--tiles-per-sheet",
+        type=int,
+        default=64,
+        help="Maximum rendered cards per contact-sheet page. Default: 64.",
+    )
     parser.add_argument(
         "--blank-range-threshold",
         type=int,
@@ -329,16 +335,17 @@ def build_report(summary: dict[str, Any], summary_path: Path, args: argparse.Nam
             }
         )
 
-    contact_sheets = []
+    if args.tiles_per_sheet <= 0:
-    if rendered_cards:
+        raise ValueError("tiles_per_sheet must be positive")
-        contact_sheets.append(
+    contact_sheets = [
-            {
+        {
-                "path": CONTACT_SHEET_NAME,
+            "path": CONTACT_SHEET_NAME_TEMPLATE.format(index=(start // args.tiles_per_sheet) + 1),
-                "tile_count": len(rendered_cards),
+            "tile_count": len(rendered_cards[start : start + args.tiles_per_sheet]),
-                "columns": args.columns,
+            "columns": args.columns,
-                "thumb_size": args.thumb_size,
+            "thumb_size": args.thumb_size,
-            }
+        }
-        )
+        for start in range(0, len(rendered_cards), args.tiles_per_sheet)
+    ]
 
     return (
         {
@@ -350,6 +357,7 @@ def build_report(summary: dict[str, Any], summary_path: Path, args: argparse.Nam
             "requested_sample_slugs": sorted(set(args.sample_slug)),
             "columns": args.columns,
             "thumb_size": args.thumb_size,
+            "tiles_per_sheet": args.tiles_per_sheet,
             "selected_tile_count": len(selected_tiles),
             "selected_sample_count": len(
                 {str(tile.get("sample_slug") or "unknown") for tile in selected_tiles}
@@ -424,8 +432,10 @@ def main() -> int:
 
     summary = load_json(summary_path)
     report, cards = build_report(summary, summary_path, args)
-    if cards:
+    for page_index, start in enumerate(range(0, len(cards), args.tiles_per_sheet), start=1):
-        build_contact_sheet(cards, args.columns, output_dir / CONTACT_SHEET_NAME)
+        page_cards = cards[start : start + args.tiles_per_sheet]
+        page_name = CONTACT_SHEET_NAME_TEMPLATE.format(index=page_index)
+        build_contact_sheet(page_cards, args.columns, output_dir / page_name)
 
     (output_dir / JSON_NAME).write_text(json.dumps(report, indent=2, sort_keys=True), encoding="utf-8")
     write_markdown(report, output_dir)

tests/test_render_operator_yolo_label_qa_contact_sheets.py

@@ -1,6 +1,9 @@
 import pytest
 
-from scripts.render_operator_yolo_label_qa_contact_sheets import filter_tiles_by_samples
+from scripts.render_operator_yolo_label_qa_contact_sheets import (
+    CONTACT_SHEET_NAME_TEMPLATE,
+    filter_tiles_by_samples,
+)
 
 
 TILES = [
@@ -24,3 +27,8 @@ def test_filter_tiles_by_samples_rejects_missing_sample() -> None:
 
 def test_filter_tiles_by_samples_without_filter_preserves_tiles() -> None:
     assert filter_tiles_by_samples(TILES, []) is TILES
+
+
+def test_contact_sheet_name_template_is_stable_and_one_indexed() -> None:
+    assert CONTACT_SHEET_NAME_TEMPLATE.format(index=1) == "contact_sheet_001.png"
+    assert CONTACT_SHEET_NAME_TEMPLATE.format(index=12) == "contact_sheet_012.png"