A Locust Recognition Method Based on Graph Convolution Neural Network

In recent years, due to frequent outbreaks of locust plagues around the world, causing catastrophic damages to a large number of food crops, locust identification is urgent. For common locust recognition methods, such as RCNN (Regions with CNN features) network training recognition, the recognition rate is lower. This paper proposes a Graph Convolutional Network(GCN)image recognition method, which is improved on the basis of ordinary convolutional neural network to train locust images to achieve higher recognition rate and ensure crops are not damaged.