Research on maize disease detection based on attention mechanism

Maize disease detection refers to the detection of maize growth images, and ultimately the detection of diseases and their sites carried by maize. At the early and middle stages of maize diseases, they are often characterized by inconspicuous pathological features and small areas of pathological sites, which constitute a great challenge for accurate maize disease detection. The attention detection mechanism determines the feature focus of the target detection model on the detection target, and the feature extraction ability of the model can be improved by feature focusing on the target, thus solving the problem of inconspicuous pathological features and small area of pathological sites to a certain extent. CBAM is a module based on the channel attention mechanism and the spatial attention mechanism, which can be effectively enhanced compared with other attention mechanisms. Therefore, in this paper, CBAM is used as an attention mechanism for maize disease detection, which focuses on three factors such as pathology color, pathology area and pathology shape of maize diseases. In this paper, this mechanism is integrated into YOLOv5, and validated on the publicly available maize pathology detection dataset, and its detection accuracy reaches 92.8%, and mAP0.5 is improved by 4% compared with the original YOLOv5 detection algorithm, which achieves good experimental results.