Table 3 Performance comparison of different object detection algorithms on M3FD and LLVIP datasets.
From: Super Mamba feature enhancement framework for small object detection
Method | M3FD | LLVIP | ||||||
|---|---|---|---|---|---|---|---|---|
mAP@0.5 | mAP@0.75 | mAP@0.95 | Pare(M) | mAP@0.5 | mAP@0.75 | mAP@0.95 | Pare(M) | |
Super-yolo-star [45] | 0.8722 | 0.8601 | 0.8437 | 5.1 | 0.8789 | 0.8063 | 0.7746 | 5.1 |
Super-yolo-Dattention [46] | 0.8653 | 0.8507 | 0.8254 | 5.3 | 0.8745 | 0.8025 | 0.7507 | 5.3 |
Super-yolo-RFAConv [41] | 0.8751 | 0.8520 | 0.8373 | 5.6 | 0.8628 | 0.7990 | 0.7206 | 5.6 |
Yolov5 [47] | 0.7983 | 0.7654 | 0.7434 | 17.7 | 0.8432 | 0.8103 | 0.7143 | 17.7 |
Yolov11 [49] | 0.8940 | 0.8736 | 0.8501 | 18.3 | 0.9145 | 0.8405 | 0.8038 | 18.3 |
Yolov8 [48] | 0.9017 | 0.8832 | 0.8478 | 21.5 | 0.9222 | 0.8316 | 0.8043 | 21.5 |
Ours | 0.9322 | 0.9034 | 0.8865 | 17.6 | 0.9413 | 0.8775 | 0.8514 | 17.6 |
