SIGNATURE WORK
CONFERENCE & EXHIBITION 2022

The management of wild animals in many safari parks has always been a man-made operation, which has led to many incidents of wild animals attacking people. In addition, stray animal problem is now a serious problem in the urban area, especially for developing cities. Stray animals might cause potential diseases, affect residents’ life quality, impact traffic safety, cause environmental problems, etc. Therefore, automated management of wild and stray animals is an important solution to these problems. This project will introduce and show how to use image recognition to contribute to the implementation of wildlife management and TNR, and study how the type and depth of neural networks affect image recognition. On the hardware side, as a small and hidden camera platform that integrates shooting, downloading, and uploading can play a vital role, Raspberry Pi is qualified for this project. The method I used in this project is transfer learning. Transfer learning is the application of knowledge or patterns learned in a certain field or task to different but related fields or problems. I choose to train and test CNN in Paddle Paddle. After I trained VGG 16, VGG 19, Mobile Net, Resnet, 50 Resnet, 101 Resnet, 152 Alex Net, I found that the best performing CNN is Mobile Net.