Identification of first-level national protected animals based on YOLOv5 and YOLOv8 algorithms

With the continuous maturity and development of computer vision technology, at present, animal protection departments in China and even in various countries around the world have spent a lot of human and material resources and financial resources on the number and distribution of wild animals and are committed to studying various wildlife detection systems to gradually re-duce the interference of artificial investigations on the habitat of wild animals. This project uses the fast image recognition technology of the realtime object detection algorithm YOLO series(including YOLOv5 and YOLOv8) and strives to build a model that can quickly and accurately identify endangered animals in complex field envi- ronments. We carefully selected 17 endangered wild animals from the list of Class I protected animals as the targets for identifica- tion, and their images were collected, labeled, and used for model training and testing. After in-depth analysis and comparison of the training results of different models, we selected the best performing model and integrated it into a visual page for easy display. During the study, we found that compared with YOLOv5, YOLOv8 had better detection effect in similar models. At the same time, we also noticed that the increase of model size will reduce the detection speed, but the corresponding detection effect will be improved. We applied this optimization model to the field detection camera to achieve fast and accurate identification of endangered protected animals. The practical application of this model will play a positive role in solving the problems such as the reduction of the number of endangered animals and the threat of ecological environment, and provide powerful technical support for the survival and repro- duction of precious creatures.