Progress in the application of deep learning in wildlife image recognition and analysis

Establishing a comprehensive wildlife monitoring system is the foundation for conducting conservation research. Traditional manual monitoring methods have various limitations, and some monitoring efforts have gradually been replaced by infrared camera trap technology. Nevertheless, the widespread use of infrared camera monitoring technology has introduced challenges in handling and analyzing massive amounts of data. Therefore, it is urgent to find an efficient method to process and analyze a large number of infrared camera data. In recent years, deep learning has been widely applied in the study of wild animal images. In order to comprehensively understand the application progress of deep learning theory and technology in wildlife image recognition, we provide an overview of the relevant research from 2000 to 2024. It elaborates on commonly used network models applications and their research progress in terms of eliminating invalid data, species identification, individual recognition, and behavior recognition. We summarize the status of deep learning in two types of images of wild animals, and emphatically discuss the existing problems and solutions of deep learning in infrared camera images. This paper analyzes the potential of applying artificial intelligence image processing techniques in infrared camera monitoring work and provides recommendations and insights for future development in order to provide ideas and directions for research on individual identification and population monitoring of wild animals.