Table 5 Results on the GC10-DET dataset compared with SOTA defect detectors.
From: Dual-branch information extraction and local attention anchor-free network for defect detection
Method | Pu | Wl | Cg | Ws | Os | Ss | In | Rp | Cr | Wf | mAP (%) |
|---|---|---|---|---|---|---|---|---|---|---|---|
DCC-CenterNet30 | 84.4 | 85.5 | 96.2 | 77.3 | 50.9 | 54.8 | 30.2 | 13.9 | 49.9 | 76.6 | 61.9 |
90.0 | 88.5 | 84.8 | 55.8 | 62.2 | 65.0 | 25.6 | 36.4 | 52.1 | 91.9 | 65.1 | |
Wang’s YOLOv361 | 62.8 | 59.7 | 85.1 | 68.6 | 27.2 | 30.7 | 8.9 | 1.5 | 16.4 | 50.3 | 41.1 |
MSC-DNet11 | 95.5 | 96.1 | 94.9 | 76.5 | 66.5 | 65.8 | 34.1 | 53.4 | 48.5 | 84.0 | 71.6 |
DLA-FCOS (Ours) | 96.5 | 92.4 | 94.6 | 81.0 | 66.8 | 65.2 | 40.1 | 27.8 | 31.9 | 77.6 | 67.7 |
