Maize Leaf Disease Identification Method Based on Improved Faster R-CNN

Maize leaf disease is a common problem in the process of crop growth. It is of great significance to identify and monitor the disease timely and accurately for the benefit and yield of agricultural production. This paper aims to propose a method of maize leaf disease recognition based on improved Faster R-CNN. First, Region Proposal Network (RPN) is introduced in Faster R-CNN to generate candidate regions to reduce the computational load in the identification process. Secondly, the network uses Adam optimizer to optimize the training process of the model to improve the convergence speed and performance. Finally, in order to better deal with multiple categories of diseases, we use the binary cross entropy loss function to classify them, and introduce the box regression loss function to accurately predict the disease boundary box. The experimental results show that the method can improve the accuracy of disease identification and recall rate significantly. It has important application value for crop disease monitoring and agricultural production.