GSD-YOLO: A Lightweight Decoupled Wheat Scab Spore Detection Network Based on Yolov7-Tiny

Aimed at the problem of the difference between intra-class and inter-class pathogenic spores of Wheat Scab image being small and difficult to distinguish, in this paper, we propose a lightweight decoupled Wheat Scab spore detection network based on Yolov7-tiny (GSD-YOLO). Specifically, considering the limitations of the storage space and power consumption of actual field detection equipment, the original detection head is optimized as a decoupled head, and the GSConv lightweight module is embedded to reduce the parameters of the model and the number of calculations required. In addition, we utilize an improved Spore–Copy data augmentation strategy to improve the detection performance and generalization ability of the algorithm to fit the large numbers, morphology, and variety of wheat disease spores in the actual field and to improve the efficiency of constructing a large data set of diverse spores. The experimental results show that the mAP of the proposed algorithm reaches 98.0%, which is 3.9 percentage points higher than that of the original model. At the same time, the detection speed of the algorithm is 114 f/s, and the memory is 13.1 MB, which meets the application requirements of hardware deployment and real-time detection. It can provide some technical support to the prevention and grading of Wheat Scab in actual farmland.