Leaf identification of apple pests and diseases based on improved ResNeSt

The occurrence of pests and diseases will seriously affect the quality and yield of fruits. The diagnosis and identification of pests and diseases is of great significance to timely correct the symptoms and improve the production quality and economic benefits. Aiming at efficient and accurate identification of apple pests and diseases, a new method based on improved ResNeS and transfer learning is proposed. Compared with ResNet, this model introduced Multi-branch structure and Split-Attention module, and introduced CA attention mechanism into Split-Attention, which effectively improved the accuracy and robustness of the network. Based on the transfer learning of ImageNet data set, the improved network finally achieves the recognition accuracy of 99.29%. This study achieved the recognition of five types of Healthy, Scab, Rust, powdery_mildew, frog_eye_leaf_spot, which is helpful to the intelligent control of apple pests and diseases.