AAL-Net: A Lightweight Detection Method for Road Surface Defects Based on Attention and Data Augmentation

The pothole is a common road defect that seriously affects traffic efficiency and personal safety. Road evaluation maintenance automatic driving take detection as their main research part. In the above scenarios, accuracy real-time are most important. However, current methods can not meet requirements of due to multiple parameters volume. To solve these problems, we first propose lightweight one-stage object network, AAL-Net. design an LF (lightweight feature extraction) module use NAM (Normalization-based Attention Module) attention ensure real time process. Secondly, make our own dataset for detection. Finally, in order simulate scene, data augmentation method further improve robustness metrics F1 GFLOPs show better than other deep learning models self-made pothole600 well