Results, losses & training diagnosis

035737_f3ac6323 · completed_with_weak_training

Scan overlap and projected CAD labels

Automatic weak mappings. Defect names follow CSV row numbers. Poor fits and rejected annotations stay visible; only eligible datasets enter experimental training. Mapping evidence and eligibility JSON

DatasetMedian fit (assumed mm)AnnotationsWeak nodesEligiblePNGs
AC28_DT_FRT_UPR_LHD_20260909_15_15__NG_a22f7d301de472.071 accepted / 2 rejected0FalseOverlap · CAD labels
AC28_DT_RR_UPR_STD_1_20260827_18_01__NG_7194c0e36c7b3.900 accepted / 1 rejected0FalseOverlap · CAD labels
AC28_DT_RR_UPR_STD_2_20260827_18_01__NG_8df93616f7ea2.101 accepted / 0 rejected222TrueOverlap · CAD labels
AC28_DT_RR_UPR_STD_20260827_16__NG_008adde420f04.511 accepted / 0 rejected129TrueOverlap · CAD labels
AC28_DT_RR_UPR_STD_20260828_09_21__NG_f4c68f38e4677.601 accepted / 0 rejected130TrueOverlap · CAD labels
LT2_DT_FRT_AREST2_20260914_15_10__NG_24389593753514.301 accepted / 1 rejected0FalseOverlap · CAD labels
ME1a_DT_FRT_UPR_LHD_20260608_14_06__NG_ffc4fce5d99614.036 accepted / 0 rejected0FalseOverlap · CAD labels
MX5a_PE_DT_RR_AREST_1_20260626_13_08__NG_fe0e1b9b8b5322.124 accepted / 0 rejected0FalseOverlap · CAD labels
MX5a_PE_DT_RR_AREST_2_20260626_13_08__NG_41ec5d7a218e17.993 accepted / 1 rejected0FalseOverlap · CAD labels
NX5_DT_RR_UPR-2_CUR_1_20260916_10_40__NG_1f98f526a60c7.743 accepted / 1 rejected0FalseOverlap · CAD labels
NX5_DT_RR_UPR-2_CUR_1_20260916_10_53__NG_b3a8f50d89ae7.382 accepted / 1 rejected0FalseOverlap · CAD labels
NX5_DT_RR_UPR-2_CUR_1_20260916_11_40__NG_1b598f48d4e16.402 accepted / 0 rejected93TrueOverlap · CAD labels
NX5_DT_RR_UPR-2_CUR_2_20260916_10_40__NG_4debd31ebf8a6.252 accepted / 2 rejected0FalseOverlap · CAD labels
NX5_DT_RR_UPR-2_CUR_2_20260916_10_53__NG_a9e70207f7eb6.951 accepted / 1 rejected0FalseOverlap · CAD labels
# Training diagnosis: 20261020 / 035737_f3ac6323

State: **completed_with_weak_training**. Trained predictions: 140; scored records: 0; folds with epoch losses: 9.

Reviewed dataset labels: 0.

- gaussian_process: requires_training_labels (14 records): GP requires a fitted training posterior; no random/synthetic prior score emitted

1. Review CAD-node risk labels against the exact geometry_sha256 and node order; use each dataset's review_template.json.
2. CSV X/Y/W/H are image pixels. Supply a reviewed pixel-to-CAD mapping before treating them as CAD labels.
3. Run main.py --prepared-manifest <saved prepared_manifest.json> --labels-dir <reviewed-label-folder> to retain the exact labeled geometry.
4. sparse_unet requires MinkowskiEngine. Choose available models with --models if that optional backend is absent.
Blocked training, backend availability and actual forward runs.
Blocked training, backend availability and actual forward runs.
Trained and reviewed results, separated by evaluation protocol. Empty when no scores exist.
Trained and reviewed results, separated by evaluation protocol. Empty when no scores exist.
Actual recorded training histories.
Actual recorded training histories.
Recorded optimizer loss: cnn2d_unet / train_all / all.
Recorded optimizer loss: cnn2d_unet / train_all / all.
Recorded optimizer loss: dgcnn / train_all / all.
Recorded optimizer loss: dgcnn / train_all / all.
Recorded optimizer loss: fno2d / train_all / all.
Recorded optimizer loss: fno2d / train_all / all.
Recorded optimizer loss: meshgraphnet / train_all / all.
Recorded optimizer loss: meshgraphnet / train_all / all.
Recorded optimizer loss: physics_risk_gnn / train_all / all.
Recorded optimizer loss: physics_risk_gnn / train_all / all.
Recorded optimizer loss: pointnet / train_all / all.
Recorded optimizer loss: pointnet / train_all / all.
Recorded optimizer loss: tabular_mlp / train_all / all.
Recorded optimizer loss: tabular_mlp / train_all / all.
Recorded optimizer loss: unet3d / train_all / all.
Recorded optimizer loss: unet3d / train_all / all.
Recorded optimizer loss: uvnet / train_all / all.
Recorded optimizer loss: uvnet / train_all / all.
Diagnostic coverage and timing describe execution, not predictive accuracy.
Diagnostic coverage and timing describe execution, not predictive accuracy.
CSV pixel annotations on the source scan image: AC28_DT_FRT_UPR_LHD_20260909_15_15__NG_a22f7d301de4. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_FRT_UPR_LHD_20260909_15_15__NG_a22f7d301de4. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_RR_UPR_STD_1_20260827_18_01__NG_7194c0e36c7b. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_RR_UPR_STD_1_20260827_18_01__NG_7194c0e36c7b. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_RR_UPR_STD_20260827_16__NG_008adde420f0. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_RR_UPR_STD_20260827_16__NG_008adde420f0. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_RR_UPR_STD_20260828_09_21__NG_f4c68f38e467. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_RR_UPR_STD_20260828_09_21__NG_f4c68f38e467. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_RR_UPR_STD_2_20260827_18_01__NG_8df93616f7ea. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: AC28_DT_RR_UPR_STD_2_20260827_18_01__NG_8df93616f7ea. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: LT2_DT_FRT_AREST2_20260914_15_10__NG_243895937535. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: LT2_DT_FRT_AREST2_20260914_15_10__NG_243895937535. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: ME1a_DT_FRT_UPR_LHD_20260608_14_06__NG_ffc4fce5d996. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: ME1a_DT_FRT_UPR_LHD_20260608_14_06__NG_ffc4fce5d996. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: MX5a_PE_DT_RR_AREST_1_20260626_13_08__NG_fe0e1b9b8b53. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: MX5a_PE_DT_RR_AREST_1_20260626_13_08__NG_fe0e1b9b8b53. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: MX5a_PE_DT_RR_AREST_2_20260626_13_08__NG_41ec5d7a218e. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: MX5a_PE_DT_RR_AREST_2_20260626_13_08__NG_41ec5d7a218e. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_1_20260916_10_40__NG_1f98f526a60c. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_1_20260916_10_40__NG_1f98f526a60c. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_1_20260916_10_53__NG_b3a8f50d89ae. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_1_20260916_10_53__NG_b3a8f50d89ae. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_1_20260916_11_40__NG_1b598f48d4e1. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_1_20260916_11_40__NG_1b598f48d4e1. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_2_20260916_10_40__NG_4debd31ebf8a. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_2_20260916_10_40__NG_4debd31ebf8a. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_2_20260916_10_53__NG_a9e70207f7eb. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
CSV pixel annotations on the source scan image: NX5_DT_RR_UPR-2_CUR_2_20260916_10_53__NG_a9e70207f7eb. Each marker shows the CSV row number followed by its Error / defect label when supplied. W/H boxes assume X/Y top-left for inspection; the anchor convention and CAD mapping are unvalidated.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_FRT_UPR_LHD_20260909_15_15__NG_a22f7d301de4. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_FRT_UPR_LHD_20260909_15_15__NG_a22f7d301de4. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_RR_UPR_STD_1_20260827_18_01__NG_7194c0e36c7b. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_RR_UPR_STD_1_20260827_18_01__NG_7194c0e36c7b. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_RR_UPR_STD_20260827_16__NG_008adde420f0. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_RR_UPR_STD_20260827_16__NG_008adde420f0. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_RR_UPR_STD_20260828_09_21__NG_f4c68f38e467. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_RR_UPR_STD_20260828_09_21__NG_f4c68f38e467. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_RR_UPR_STD_2_20260827_18_01__NG_8df93616f7ea. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: AC28_DT_RR_UPR_STD_2_20260827_18_01__NG_8df93616f7ea. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: LT2_DT_FRT_AREST2_20260914_15_10__NG_243895937535. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: LT2_DT_FRT_AREST2_20260914_15_10__NG_243895937535. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: ME1a_DT_FRT_UPR_LHD_20260608_14_06__NG_ffc4fce5d996. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: ME1a_DT_FRT_UPR_LHD_20260608_14_06__NG_ffc4fce5d996. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: MX5a_PE_DT_RR_AREST_1_20260626_13_08__NG_fe0e1b9b8b53. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: MX5a_PE_DT_RR_AREST_1_20260626_13_08__NG_fe0e1b9b8b53. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: MX5a_PE_DT_RR_AREST_2_20260626_13_08__NG_41ec5d7a218e. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: MX5a_PE_DT_RR_AREST_2_20260626_13_08__NG_41ec5d7a218e. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_1_20260916_10_40__NG_1f98f526a60c. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_1_20260916_10_40__NG_1f98f526a60c. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_1_20260916_10_53__NG_b3a8f50d89ae. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_1_20260916_10_53__NG_b3a8f50d89ae. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_1_20260916_11_40__NG_1b598f48d4e1. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_1_20260916_11_40__NG_1b598f48d4e1. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_2_20260916_10_40__NG_4debd31ebf8a. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_2_20260916_10_40__NG_4debd31ebf8a. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_2_20260916_10_53__NG_a9e70207f7eb. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
Fused scan rigidly overlapped with CAD in three views: NX5_DT_RR_UPR-2_CUR_2_20260916_10_53__NG_a9e70207f7eb. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_FRT_UPR_LHD_20260909_15_15__NG_a22f7d301de4. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_FRT_UPR_LHD_20260909_15_15__NG_a22f7d301de4. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_RR_UPR_STD_1_20260827_18_01__NG_7194c0e36c7b. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_RR_UPR_STD_1_20260827_18_01__NG_7194c0e36c7b. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_RR_UPR_STD_20260827_16__NG_008adde420f0. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_RR_UPR_STD_20260827_16__NG_008adde420f0. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_RR_UPR_STD_20260828_09_21__NG_f4c68f38e467. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_RR_UPR_STD_20260828_09_21__NG_f4c68f38e467. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_RR_UPR_STD_2_20260827_18_01__NG_8df93616f7ea. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: AC28_DT_RR_UPR_STD_2_20260827_18_01__NG_8df93616f7ea. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: LT2_DT_FRT_AREST2_20260914_15_10__NG_243895937535. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: LT2_DT_FRT_AREST2_20260914_15_10__NG_243895937535. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: ME1a_DT_FRT_UPR_LHD_20260608_14_06__NG_ffc4fce5d996. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: ME1a_DT_FRT_UPR_LHD_20260608_14_06__NG_ffc4fce5d996. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: MX5a_PE_DT_RR_AREST_1_20260626_13_08__NG_fe0e1b9b8b53. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: MX5a_PE_DT_RR_AREST_1_20260626_13_08__NG_fe0e1b9b8b53. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: MX5a_PE_DT_RR_AREST_2_20260626_13_08__NG_41ec5d7a218e. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: MX5a_PE_DT_RR_AREST_2_20260626_13_08__NG_41ec5d7a218e. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_1_20260916_10_40__NG_1f98f526a60c. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_1_20260916_10_40__NG_1f98f526a60c. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_1_20260916_10_53__NG_b3a8f50d89ae. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_1_20260916_10_53__NG_b3a8f50d89ae. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_1_20260916_11_40__NG_1b598f48d4e1. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_1_20260916_11_40__NG_1b598f48d4e1. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_2_20260916_10_40__NG_4debd31ebf8a. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_2_20260916_10_40__NG_4debd31ebf8a. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_2_20260916_10_53__NG_a9e70207f7eb. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.
CSV defect names and regions projected onto CAD; exact model-node weak labels: NX5_DT_RR_UPR-2_CUR_2_20260916_10_53__NG_a9e70207f7eb. Automatic geometric mapping; rejected candidates and weak-label eligibility are shown. This is not independently reviewed ground truth.