The dataset of fish surface pathology was constructed based on four types of fish diseases with high rate of incidence and great harm to fish to improve the accuracy and efficiency of identifying fish surface pathology and solve the problems of heavy reliance on manual labor and low accuracy of identification in the process of identification at present. An improved and optimized DBA_Resnet-18 model with high accuracy of identification based on the Resnet-18 model was constructed by integrating spatial attention and SE channel attention dual attention mechanism. A real-time intelligent visualization system for identifying fish diseases was developed based on this model as well. The improved model incorporates SE channel attention module in the middle of the network and introduces spatial attention mechanism at the end of the network. The results of testing showed that the accuracy of the DBA_Resnet-18 model in classifying fish surface pathology reached 96.75%， which was 1.71， 2.12， 2.37， 2.83， 2.51， 2.23， 2.50， and 3.53 percent points higher than that of the commonly used models including Resnet-18， Resnet-34， Resnet-50， Resnet-101， Swin Transformer， VGG-16， VGG-19， and AlexNet， respectively. It is indicated that the proposed model and the developed intelligent visualization system for identifying fish diseases can quickly and accurately classify and identify different fish surface pathologies， realizing the intelligence of the system for identifying fish diseases， which can be used to diagnose the types of fish surface pathology in practical environments.