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 2 改进前、后模型各指标数据对比 |
| Tab.2 Performance metrics of original and improved models |
|
| 病害类别 | P1/% | P2/% | R1/% | R2/% | mAP@0.51/% | mAP@0.52/% | mAP@0.5:0.951/% | mAP@0.5:0.952/% | | 注：X1表示YOLO12-n相关指标，X2表示DKP-YOLO相关指标 | | 所有类别 | 78.0 | 90.4 (+12.4) | 71.7 | 80.4 (+8.7) | 76.0 | 84.9 (+8.9) | 48.3 | 61.9 (+13.6) | | 网状裂缝 | 82.7 | 93.5 (+10.8) | 71.4 | 91.6 (+20.2) | 79.6 | 93.7 (+14.1) | 51.5 | 68.4 (+16.9) | | 纵向裂缝 | 81.7 | 95.0 (+13.3) | 74.5 | 81.4 (+6.9) | 79.4 | 89.9 (+10.5) | 48.2 | 62.9 (+14.7) | | 斜向裂缝 | 72.5 | 92.9 (+20.4) | 62.9 | 74.3 (+11.4) | 66.4 | 82.6 (+16.2) | 39.6 | 56.3 (+16.7) | | 坑洞 | 74.3 | 85.4 (+11.1) | 60.0 | 58.5 (−1.5) | 57.5 | 60.1 (+2.6) | 33.0 | 46.0 (+13.0) | | 修补 | 72.6 | 80.3 (+7.7) | 82.1 | 92.9 (+10.8) | 88.3 | 92.8 (+4.5) | 63.5 | 72.3 (+8.8) | | 横向裂缝 | 84.6 | 95.4 (+10.8) | 79.5 | 83.8 (+4.3) | 84.9 | 90.7 (+5.8) | 54.1 | 65.6 (+11.5) |
|
|
|