Research on identification of main cotton pests based on deep learning

Abstract

According to the phototaxis of cotton bollworm adults, the self-developed cotton bollworm adult trapping and photographing monitoring device automatically obtains images of cotton bollworm adults; the target detection algorithm YOLO v5 is used to identify and count the cotton bollworm adults on the monitoring equipment; comparison The detection performance of different training models on images of H. armigera adult images was evaluated, and the differences of each model were evaluated by precision rate, recall rate, F1 value and average precision. The test results show that high recognition accuracy can be achieved when using the monitor to collect images of the test set as the training set.. It can improve the current situation of low automation of cotton bollworm adult identification, and can be used for the actual field monitoring of cotton bollworm adults.