基于动态核感知的无人机视角路面病害检测方法
|
|
张建刚,李肖,冯丹丹
|
Dynamic kernel perception for pavement distress detection in UAV inspection
|
|
Jiangang ZHANG,Xiao LI,Dandan FENG
|
|
| 表 4 各模型检测性能评价指标 |
| Tab.4 Evaluation metrics for detection performance of different models |
|
| 模型 | P/% | R/% | mAP@0.5:0.95/% | F1/% | FPS/(帧·s−1) | Params/106 | GFLOPs/109 | Weights/MB | | 1）注：加粗字体表示每列中的最优值 | | SSD | 69.2 | 54.4 | 42.5 | 60.9 | 40.0 | 32.3 | 128.9 | 49.1 | | Faster-RCNN | 73.9 | 59.2 | 46.3 | 65.7 | 14.0 | 136.8 | 370.2 | 108.2 | | RT-DETR | 70.3 | 56.4 | 34.1 | 62.6 | 77.0 | 4.2 | 130.5 | 83.0 | | UAV-YOLO | 84.3 | 69.2 | 48.6 | 76.0 | 78.0 | 10.5 | 25.3 | 19.4 | | LFDS-YOLO | 90.4 | 75.7 | 53.0 | 82.4 | 95.0 | 4.4 | 20.2 | 8.7 | | YOLOv5-n | 82.1 | 71.4 | 50.0 | 76.3 | 257.9 | 2.5 | 7.2 | 5.3 | | YOLOv6-n | 72.2 | 56.1 | 32.6 | 63.1 | 253.2 | 4.2 | 11.9 | 8.4 | | YOLOv8-n | 90.3 | 74.1 | 54.3 | 81.4 | 263.8 | 3.0 | 8.2 | 6.3 | | YOLOv9-s | 90.61） | 75.1 | 57.6 | 82.1 | 196.0 | 7.3 | 27.4 | 15.3 | | YOLOv9-t | 82.2 | 66.0 | 44.4 | 73.2 | 248.8 | 2.1 | 7.9 | 4.7 | | YOLOv10-n | 86.2 | 74.1 | 55.0 | 79.7 | 261.1 | 2.7 | 8.4 | 5.6 | | YOLO11-n | 90.6 | 77.9 | 56.5 | 83.7 | 256.3 | 2.6 | 6.4 | 5.3 | | YOLO12-n | 78.0 | 71.7 | 48.3 | 74.7 | 212.7 | 2.6 | 6.2 | 5.3 | | YOLO13-n | 82.6 | 64.7 | 44.9 | 72.6 | 191.5 | 2.5 | 6.4 | 5.4 | | DKP-YOLO | 90.4 | 80.4 | 61.9 | 85.1 | 225.8 | 2.4 | 5.5 | 5.2 |
|
|
|