Analysis and Research on Plant Disease Identification Based on Swin Transformer by Embedding Convolutional Block Attention Module

In recent years, climate change and changes in agricultural activities have significantly increased the frequency and severity of plant diseases, with significant implications for food production and quality security. Therefore, timely and accurate detection and diagnosis of plant diseases is crucial for food security. To address this challenge, a Swin Transformer model embedded with an attention mechanism is designed in this paper. The model uses a Swin Transformer as the back-bone network and embeds a CBAM hybrid attention module. CBAM combines the channel attention mechanism and the spatial attention mechanism and can adaptively assign higher weights to essential features and suppress irrelevant or redundant features, which helps the model to more accurately capture the key features of plant leaf diseases under complex backgrounds and improve recognition accuracy. In addition, the model adds a Layer Norm layer, an adaptive average pooling layer, a fully connected layer, and a Softmax layer to the top layer of the structure to optimize the feature extraction and classification performance further. Experimental results show that the proposed model improves accuracy on multiple datasets: 99.90% on VillagePlant, 90.50% on PlantDoc. The model exhibits higher accuracy and robustness compared to traditional convolutional neural networks and visual Transformer-based models.