Crack Detection on Artwork Crack (test)
87.44AccuracySegFormer
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SegFormerGuided Logit=false, LoRA=false, Mask Loss=false2026.02 | 87.44 | 27.64 | 16.04 | 28.54 | 25.82 | |
| SegFormerAblated Components=None, Filling method=AD2026.02 | 87.44 | 27.64 | — | — | — | |
| Eq. (3)Guided Logit=true, LoRA=true, Mask Loss=true2026.02 | 85.19 | 61.36 | 43.86 | 60.97 | 58.75 | |
| Learning-based SegmentationDetection method=Learning-based, Filling method=AD2026.02 | 85.19 | 61.36 | — | — | — | |
| Eq. (3)Guided Logit=true, LoRA=true, Mask Loss=true, Filling method=AD2026.02 | 85.19 | 61.36 | — | — | — | |
| Eq. (4)Guided Logit=true, LoRA=false, Mask Loss=false2026.02 | 85.1 | 28.3 | 26.5 | 35.3 | 33.6 | |
| Eq. (4)Guided Logit=true, Filling method=AD2026.02 | 85.1 | 28.3 | — | — | — | |
| Eq. (1)Guided Logit=true, LoRA=true, Mask Loss=false2026.02 | 82.79 | 49.45 | 32.78 | 49.37 | 41.69 | |
| Eq. (1)Guided Logit=true, LoRA=true, Filling method=AD2026.02 | 82.79 | 49.45 | — | — | — | |
| Learning-based SegmentationDetection method=Learning-based, Filling method=MTM2026.02 | 82.3 | 59.65 | — | — | — | |
| Grassfire AlgorithmDetection method=Grassfire, Filling method=AD2026.02 | 74.11 | 51.36 | — | — | — | |
| Grassfire AlgorithmDetection method=Grassfire, Filling method=MTM2026.02 | 71.75 | 49.59 | — | — | — |