DSC-T-Yolo-Rice: A Sand Clock Yolo Model for Rice Leaves Diseases Detection

In this study, a new Deep Sand Clock Tiny-Yolo Rice (DSC-T-yolo-Rice) network is proposed for rice leaf disease detection. Based on Tiny Yolo v4 network, the following modules are accomplished to improve the accuracy such as the Spatial Pyramid Pooling module (SPP), Convolutional Block Attention Module (CBAM), and Deep Sand Clock Feature Extraction Module (DSCFEM). Through the experimental results, it is proved that the proposed method achieves the highest test mAP.In addition, the proposed architecture in this paper focuses on the detection and analysis of key features.