Table 7 Comprehensive efficiency-accuracy trade-off (vs. YOLOv11s: NEU-DET +7.3% \(\hbox {mAP}_{50}\), +151% FPS; GC10-DET +11.6% \(\hbox {mAP}_{50}\), +56% FPS).
Datasets | Methods | Params | Gflops | FPS | \(\hbox {mAP}_{50}\) | \(\hbox {mAP}_{50-95}\) |
|---|---|---|---|---|---|---|
NEU-DET | Faster R-CNN16 | 41.364 | 90.913 | 1.0 | 65.4 | 34.8 |
YOLOv5s | 9.13 | 192.67 | 13.42 | 76.9 | 43.4 | |
YOLOv7 | 43.5 | 130.2 | 3.797 | 70.5 | 40.3 | |
YOLOv8s | 11.14 | 229.5 | 52 | 75.2 | 43 | |
YOLOv9s | 7.29 | 219.17 | 16.26 | 77.6 | 44.5 | |
YOLOv10s | 8.07 | 198.33 | 42.36 | 70.7 | 40.8 | |
YOLOv11s | 10.73 | 213.06 | 62.02 | 75 | 43.1 | |
DAB-DETR17 | 44 | 216 | 18 | 68.6 | 33.9 | |
Deformable-DETR18 | 40 | 173 | 18 | 64.3 | - | |
DINO19 | 218 | - | 14 | 55.2 | 43 | |
LFF-YOLO20 | 60.51 | 6.85 | 42 | 73.1 | - | |
FFDDNet15 | 10.08 | 222.51 | 38 | 76.8 | 44.4 | |
LE-YOLOv521 | 4.8 | 10.3 | 55.1 | 79.1 | 41.0 | |
Ours | 8.46 | 159 | 156.1 | 80.5 | 46.9 | |
GC10-DET | Faster R-CNN16 | 41.364 | 90.913 | 27 | 55.4 | / |
YOLOv5s | 9.13 | 192.67 | 60.35 | 61.7 | 32 | |
YOLOv7 | 43.5 | 130.2 | 35 | 58.0 | / | |
YOLOv8s | 11.14 | 229.5 | 48 | 61.3 | 31.4 | |
YOLOv9s | 7.29 | 219.17 | 45 | 62.9 | 32.1 | |
YOLOv10s | 8.07 | 198.33 | 88.54 | 55.4 | 28.8 | |
YOLOv11s | 10.73 | 213.06 | 123.23 | 60.4 | 29.7 | |
DAB-DETR17 | 44 | 216 | 18 | 52.6 | - | |
Deformable-DETR18 | 40 | 173 | 16 | 54.3 | - | |
DINO19 | 218 | - | 13 | 54.3 | - | |
LFF-YOLO20 | 60.51 | 6.85 | 40 | 41.3 | - | |
FFDDNet15 | 10.08 | 222.51 | 35 | 64.2 | 32.8 | |
LE-YOLOv521 | 4.8 | 10.3 | - | 63.8 | 28.9 | |
Ours | 8.46 | 159 | 192.2 | 67.4 | 34.3 |
