Traditional image processing has many problems in crop disease recognition, such as complicated and inefficient manual design. This paper studies the performance of deep learning algorithms in crop diseases. The research mainly optimizes the convolutional neural network from two aspects, one is the model structure, and the other is training label optimization. The basic convolutional neural network uses ResNet-50, InceptionV3, MobileNetV2, etc. Model training uses transfer learning to compare experimental data based on the performance of each m...
