Wildlife Monitoring and Identification based on Faster R-CNN
Chenxi Deng, Guoxiong Zhou, Yiqing Cai · 2023
China is a powerful country with vast territory and abundant animal resources. At present, there are more than 400 kinds of national protected animals, and there are 1999 man-made reserves. Wildlife resources have important strategic significance. Real time detection and identification of wildlife is the main work of managers. Based on the research background of wildlife image detection, this paper analyzes and summarizes the traditional image detection methods, and proposes a wildlife detection and automatic recognition method based on fast r-cnn. In this paper, the tensorflow framework is downloaded and configured, and the collected wildlife images are processed and annotated. Secondly, the preprocessed image is reconstructed according to the format of voc2007 data set, and then the wildlife image is detected and recognized by using GPU based fast r-cnn framework. The experimental results show that this method can achieve fast detection and accurate recognition of wildlife.