Table 7 Performance comparison on the small object subset. Metrics for the overall test set are included for reference.
From: Balancing complexity and accuracy for defect detection on filters with an improved RT-DETR
Model | mAP@0.5 (%) | Recall (%) | ||
|---|---|---|---|---|
Overall | Small objects | Overall | Small objects | |
RT-DETR (baseline) | 90.32 | 87.14 | 90.54 | 85.71 |
RT-DETR-LGA (ours) | 97.67 | 95.93 | 97.07 | 95.21 |
Performance gain | +7.35 | +8.79 | +6.53 | +9.5 |
