Research on Classification and Detection of the Asphalt Pavement Diseases Based on Swin Transformer
Aiming at the problems of insufficient ability of identifying long-distance crack structure and accuracy limitation of the asphalt pavement disease detection in traditional convolutional neural network models,the Swin Transformer model is introduced to study the classification of the asphalt pavement disease.Firstly,for the problem of low contrast of the asphalt pavement scanning im-age collected by the road inspection vehicle,the histogram equalization technology is used to process the image,and increase the image visualization effect.Secondly,three classic convolutional neural network models are selected as comparison models,and the methods of replacing the loss function and adjusting the pre-training model are used to solve the over-fitting problem during the training process.And the accuracy rate,recall rate,F1-score are selected as the evaluation index.In the final experimental results,the recog-nition accuracy of the Swin Transformer reaches 80.6％,with the F1-score of 0.776,which not only surpass the traditional convolu-tional neural network(CNN)model in overall classification accuracy,but also has a higher recognition of diseases with long-distance characteristic structures and good reliability.
万方数据
维普
