Complex Small Target Image Recognition Algorithm Based on Data Enhancement in YOLOv7

The recognition of small and complex target image is a key problem in the scientific research field, which has very important scientific value. Although FASTER-RCNN and YOLOV5 can recognize complex targets, the recognition accuracy is not high due to the difference of network structure and the application of activation function. For the samples of six kinds of insects, the scene is complex, most of the targets overlap each other, and the recognition is difficult. Firstly, this article preprocesses the image using image data augmentation technology and uses YOLOv7 for image recognition. The RELU activation function is used, with an accuracy of 0.903, Recall reaches 0.864, and mAP reaches 0.929. After comparing this algorithm with FASTER-RNN, the results show that YOLOV7 has better recognition performance and performance.