{
  "claim_boundary": "Reference implementation regression evidence; not production model accuracy.",
  "content_sha256": "aeb0f6510352f4e205e4fc4e8c423110ca72c2142016c7af846d758865f258c6",
  "implementation": "backend/app/services/qa_service.py via scripts/run_golden_qa_benchmark.py",
  "persistence": {
    "commit_count": 4,
    "metric_count": 24,
    "metric_keys": [
      "f1",
      "false_negative_count",
      "false_positive_count",
      "mean_iou",
      "precision",
      "recall"
    ],
    "quality_check_count": 4
  },
  "scenario_count": 4,
  "scenarios": [
    {
      "benchmark_id": "golden-buildings-partial-match-v1",
      "description": "Two reference building polygons and two candidate building polygons: one candidate matches ref-1, one candidate is a false positive, and ref-2 is a false negative.",
      "fixtures": {
        "candidate": "fixtures/golden/predicted_buildings.geojson",
        "reference": "fixtures/golden/reference_buildings.geojson"
      },
      "metrics": {
        "f1": 0.5,
        "false_negative_count": 1,
        "false_positive_count": 1,
        "mean_iou": 0.8339768339652592,
        "precision": 0.5,
        "recall": 0.5
      },
      "persistence": {
        "commit_count": 1,
        "metric_count": 6,
        "metric_keys": [
          "precision",
          "recall",
          "f1",
          "mean_iou",
          "false_positive_count",
          "false_negative_count"
        ],
        "quality_check_count": 1
      },
      "result_counts": {
        "candidate_feature_count": 2,
        "matches": 1,
        "reference_feature_count": 2
      },
      "status": "passed"
    },
    {
      "benchmark_id": "golden-buildings-perfect-match-v1",
      "description": "One candidate building polygon exactly matches one reference polygon.",
      "fixtures": {
        "candidate": "fixtures/golden/predicted_buildings_perfect.geojson",
        "reference": "fixtures/golden/reference_buildings_perfect.geojson"
      },
      "metrics": {
        "f1": 1.0,
        "false_negative_count": 0,
        "false_positive_count": 0,
        "mean_iou": 1.0,
        "precision": 1.0,
        "recall": 1.0
      },
      "persistence": {
        "commit_count": 1,
        "metric_count": 6,
        "metric_keys": [
          "precision",
          "recall",
          "f1",
          "mean_iou",
          "false_positive_count",
          "false_negative_count"
        ],
        "quality_check_count": 1
      },
      "result_counts": {
        "candidate_feature_count": 1,
        "matches": 1,
        "reference_feature_count": 1
      },
      "status": "passed"
    },
    {
      "benchmark_id": "golden-buildings-no-overlap-v1",
      "description": "One candidate polygon and one reference polygon do not overlap.",
      "fixtures": {
        "candidate": "fixtures/golden/predicted_buildings_no_overlap.geojson",
        "reference": "fixtures/golden/reference_buildings_no_overlap.geojson"
      },
      "metrics": {
        "f1": null,
        "false_negative_count": 1,
        "false_positive_count": 1,
        "mean_iou": null,
        "precision": 0.0,
        "recall": 0.0
      },
      "persistence": {
        "commit_count": 1,
        "metric_count": 6,
        "metric_keys": [
          "precision",
          "recall",
          "f1",
          "mean_iou",
          "false_positive_count",
          "false_negative_count"
        ],
        "quality_check_count": 1
      },
      "result_counts": {
        "candidate_feature_count": 1,
        "matches": 0,
        "reference_feature_count": 1
      },
      "status": "passed"
    },
    {
      "benchmark_id": "golden-buildings-multipolygon-match-v1",
      "description": "One candidate MultiPolygon exactly matches one reference MultiPolygon.",
      "fixtures": {
        "candidate": "fixtures/golden/predicted_buildings_multipolygon.geojson",
        "reference": "fixtures/golden/reference_buildings_multipolygon.geojson"
      },
      "metrics": {
        "f1": 1.0,
        "false_negative_count": 0,
        "false_positive_count": 0,
        "mean_iou": 1.0,
        "precision": 1.0,
        "recall": 1.0
      },
      "persistence": {
        "commit_count": 1,
        "metric_count": 6,
        "metric_keys": [
          "precision",
          "recall",
          "f1",
          "mean_iou",
          "false_positive_count",
          "false_negative_count"
        ],
        "quality_check_count": 1
      },
      "result_counts": {
        "candidate_feature_count": 1,
        "matches": 1,
        "reference_feature_count": 1
      },
      "status": "passed"
    }
  ],
  "status": "passed",
  "version": 1
}