Balance positives during precision correction

2026-07-27 14:22:30 +02:00
parent 9e2dde2b19
commit 100d9e220b
2 changed files with 43 additions and 1 deletions
@@ -91,3 +91,37 @@ def test_sampling_can_use_calibration_before_test_is_opened() -> None:
    )
    assert len(paths) == 3
    assert metadata["failure_evidence_source"] == "calibration"
+
+
+def test_precision_correction_can_balance_positive_and_negative_tiles() -> None:
+    manifest = {"samples": [{"sample_slug": "train-fl", "split": "train", "region": "flanders"}]}
+    summary = {
+        "tiles": [
+            {"sample_slug": "train-fl", "split": "train", "label_count": 2, "image_path": "/tmp/fl-pos.png"},
+            {"sample_slug": "train-fl", "split": "train", "label_count": 0, "image_path": "/tmp/fl-neg.png"},
+        ]
+    }
+    assessment = {
+        "status": "continue_training_loop",
+        "gates": {
+            "min_region_f1": 0.45,
+            "min_region_precision": 0.5,
+            "min_region_recall": 0.4,
+            "max_pure_empty_false_positives": 0,
+        },
+        "calibration": {
+            "regions": {"flanders": {"f1": 0.46, "precision": 0.45, "recall": 0.46}}
+        },
+    }
+
+    paths, metadata = MODULE.build_sampling(
+        summary=summary,
+        manifest=manifest,
+        assessment=assessment,
+        precision_positive_repeat=2,
+        negative_repeat=3,
+    )
+
+    assert paths.count(str(Path("/tmp/fl-pos.png").resolve())) == 2
+    assert paths.count(str(Path("/tmp/fl-neg.png").resolve())) == 3
+    assert metadata["precision_positive_repeat"] == 2