基于深度学习的隧道衬砌多病害检测算法
|
宋娟,贺龙喜,龙会平
|
Deep learning-based algorithm for multi defect detection in tunnel lining
|
Juan SONG,Longxi HE,Huiping LONG
|
|
表 4 网络消融实验结果表 |
Tab.4 Network ablation experiment results |
|
网络 | MobileViT | CA | TP Block | F1/% | mAP/% | v/(帧·s−1) | MS/106 | YOLOv7 | × | × | × | 71.29 | 71.13 | 50.81 | 142.3 | YOLOv7+MobileViT | √ | × | × | 73.49 | 74.81 | 51.39 | 113.9 | YOLOv7+CA | × | √ | × | 71.55 | 71.71 | 51.43 | 142.6 | YOLOv7+TP Block | × | × | √ | 72.42 | 72.53 | 51.26 | 142.6 | YOLOv7+MobileViT+CA | √ | √ | × | 74.61 | 75.72 | 52.50 | 114.2 | YOLOv7+MobileViT+ TP Block | √ | × | √ | 75.16 | 75.91 | 52.79 | 114.2 | YOLOv7+CA+ TP Block | × | √ | √ | 74.05 | 73.60 | 52.44 | 142.9 | YOLOv7+MobileViT+CA+ TP Block | √ | √ | √ | 77.43 | 77.52 | 53.86 | 114.5 |
|
|
|