Classification and detection algorithm of human skin diseases based on improved YOLOv8
In view of the problems of missing detection and high false detection rate in the current human skin disease detec-tion,YOLOv8 is used as the benchmark model to improve.Firstly,the ability of the network to process irregular data is improved by adding DSC(dysnakeconv).Secondly,the global attention module GAM is used to enhance the attention to useful image informa-tion.Finally,with the help of wise IoU boundary box loss function based on dynamic non monotonic focusing mechanism,the bal-ance of different quality anchor boxes is achieved.The experiment shows that the improved YOLOv8 model has improved the aver-age accuracy(mAP)by 2.6%and the accuracy(Precision)by 3.2%,reaching 88.2%and 89.5%respectively in the data set of hu-man common skin diseases.
万方数据
