Table 2 Comparative experiments with prior works on our dataset.
From: River floating object detection with transformer model in real time
Model | Parameters | GFLOPs | Recall (%) | Precision (%) | mAP@0.5 (%) | mAP@0.5:0.95 (%) |
|---|---|---|---|---|---|---|
SSD | 24.7Â M | 350.7 | 72.6 | 72.4 | 70.7 | 41.0 |
Faster R-CNN | 41.2Â M | 223.4 | 76.7 | 57.1 | 70.6 | 41.4 |
YOLOv5m | 21.2Â M | 49.0 | 71.9 | 70.9 | 71.8 | 45.0 |
YOLOv6s | 18.5Â M | 45.8 | 70.7 | 69.1 | 71.1 | 44.8 |
YOLOv8m | 25.9Â M | 78.9 | 71.0 | 71.1 | 72.1 | 46.5 |
YOLOv10m | 16.7Â M | 63.5 | 69.0 | 70.7 | 70.1 | 45.0 |
DETR | 41.3Â M | 97.1 | 81.2 | 50.3 | 68.7 | 39.0 |
Deformable DETR | 39.9Â M | 200.9 | 70.7 | 69.8 | 62.2 | 33.3 |
RT-DETR | 19.0Â M | 57.0 | 68.0 | 73.8 | 71.0 | 47.1 |
Our method | 14.1Â M | 44.0 | 73.7 | 71.9 | 76.0 | 47.7 |
