Improved YOLOv8-CBAM Enhanced Pothole Road Recognition-Based Model

With the rapid development of Intelligent Transport Systems, pavement condition monitoring plays an increasingly important role in the field of road maintenance and road safety. Especially in the automatic identification and repair of potholes, the traditional method often relies on manual inspection, which is not only inefficient but also easy to miss the inspection. To improve the accuracy and robustness of pothole road identification, this paper improves the YOLOv8 model and introduces the CBAM (Convolutional Block Attention Module) module. To focus on important information, this paper divides the CBAM module into two parts: spatial attention mechanism and channel attention mechanism, in which the spatial attention mechanism uses the Recurrent Neural Network model to determine the focus of attention; and the channel attention mechanism uses the Squeeze-and-Excitation Networks model to determine the attention channel. The YOLOv8-CBAM Enhanced model is established by combining the YOLOv8 model and the CBAM module. Experiments show that the recognition accuracy reaches 99.29% on the validation set, and the algorithm YOLOv8-CBAM Enhanced model in this paper improves the accuracy by 5.79 % over YOLOv8, and at the same time, it takes only 25ms to complete the prediction of an image. The YOLOv8-CBAM Enhanced model has 99.29% confusion matrix accuracy on the training set. The advantages of the YOLOv8-CBAM Enhanced model are fully demonstrated. Compared with other models, the algorithm also achieves better detection results and can effectively accomplish the road pothole detection task, which fully demonstrates the advantages of the algorithm.