Research on corn leaf disease recognition method based on deep learning
Timely identification and control of maize leaf diseases is of great significance to ensure maize yield and quality.A method of maize leaf disease recognition based on deep learning was proposed and the corresponding mobile application was de-veloped.Nine deep learning models were used:YOLOv5s,YOLOv5m,YOLOv5l,ResNet18,ResNet34,ResNet50,Efficient-Net_b0,EfficientNet_b1,and EfficientNet_b2.The results showed that the EfficientNet_b0 model had the highest balance accu-racy(96.3%)and low model complexity,and the performance of recognition accuracy,recognition speed and model size were com-pared among the four types of maize leaf disease images,including common rust,blight,gray leaf spot and healthy leaf.The mobile application on Android platform was developed based on the EfficientNet_b0 model,which could realize real-time identification of maize leaf diseases or recognition of loaded photo albums,and provide corresponding category names and confidence levels,dis-ease introduction and suggested prevention and control measures,providing a convenient,efficient and intelligent maize leaf dis-ease diagnosis tool for maize growers.It is helpful to improve the efficiency and quality of corn production.
万方数据
维普
