MobileNetV3-AP: Integrating Attention and Pathology Mechanisms for Rice Disease Classification

To solve the problems of low accuracy and complex structure of the model in the task of rice disease classification, this study proposes a new architecture, MobilenetV3-AP, based on the improvement of MobileNetV3. By incorporating a pathological enhancement module and spatial attention mechanism, the model's capability to identify and perceive rice disease characteristics is significantly enhanced. The experimental results show that MobileNetV3-AP achieves 95.37% accuracy and 95.23 % precision respectively, and the model volume is only 16.27 MB, which is much smaller than other models, demonstrating excellent performances. This study offers valuable insights and references for how to employ lightweight models to achieve efficient classification and recognition of rice diseases.