CROP DISEASES IMAGE RECOGNITION BASED ON TRANSFER LEARNING WITH CONVOLUTIONAL NEURAL NETWORK

Aiming at the problem of low recognition rate of crop disease in complex background, a method of crop disease image recognition based on convolutional neural network based on transfer learning was proposed. Firstly, the crop images were preprocessed, including image clipping, normalization and image enhancement, in order to obtain more useful information of target recognition task. Then, the AlexNet feature extraction layer in Faster R-CNN network is improved by using transfer learning to fine tune the AlexNet network, so as to solve the problem of low recognition accuracy caused by complex environment such as sunny day, cloudy day and night light supplement. Finally, based on the Faster R-CNN model, a new full connection layer module is designed, and the convolution layer trained by Faster R-CNN model in the image data set is transferred to the proposed model, and the accurate crop disease recognition results are obtained through model training. The results show that transfer learning can significantly improve the convergence speed and recognition ability of the model, and the recognition accuracy is higher than other comparison methods.