Research on Cotton Disease Identification Method Based on Deep Learning

For cotton disease detection, traditional manual methods often require inspectors to have some identification experience and are very time-consuming and labor-intensive. There is an urgent need to find a more accurate and efficient test at this time. Deep learning is an efficient method for image recognition. In this paper, we employ three deep learning models: InceptionV3, ResNet152V2, and VGG16. Each model is trained separately on the cotton image dataset, and image classification experiments are conducted. In the experiments, based on different training set to test set ratios, the recognition accuracy of all three models is above 90%, with ResNet152V2 and VGG16 having slightly higher recognition accuracy than InceptionV3. According to the experimental results, deep learning methods successfully detect cotton diseases, providing technological support to enhance cotton yields.