I. Introduction
In the era of achieving long-term sustainable development of human society, ecological environment protection should be regarded as an important subject, and active efforts should be made to strengthen research, exploration and practice in this field. Accurate identification of these bird species is crucial for understanding their distribution, behavior, and population dynamics. Traditional methods of bird identification heavily rely on manual observation and field expertise, which are not only time-consuming but also prone to subjective factors. This limitation hinders efficient development and accurate identification of large-scale bird data. However, recent advancements in deep learning technology had opened up new possibilities for efficient and reliable image recognition, offering a promising avenue for the study of local bird species in Yunnan province.