Table 1 Ablation study on YOLO-LRDD
From: YOLO-LRDD: a lightweight method for road damage detection based on improved YOLOv5s
Method | Precision (%) | #Param. (M) | F1-Score (%) | Inference time per image (ms) | ||
|---|---|---|---|---|---|---|
Backbone | Neck | Loss | ||||
CSPdarknet + SPPF | PANet | CIOU | 56.8 | 27.6 | 57.5 | 4.4 |
ShuffleNetV2 | PANet | CIOU | 55.4 | 15.2 | 55.6 | 2.9 |
ShuffleNetV2 | BiFPN | CIOU | 55.9 | 18.5 | 56.5 | 3.1 |
Shuffle-ECANet | BiFPN | CIOU | 56.8 | 19.6 | 57.1 | 3.2 |
Shuffle-ECANet | BiFPN | Focal-EIOU | 57.1 | 19.8 | 58.7 | 3.4 |