With the rapid growth of China's electrified railway mileage, the safe and stable operation of catenary is facing tremendous pressure, so it is of great significance to monitor. In this paper, the avian hazards affecting the normal operation of the catenary of electrified railway were studied. Through the analysis and comparison of different depth learning models, DSSD model was selected to automatically identify the catenary of high-speed railway. At the same time, this paper used transfer learning method and Caffe platform to improve the generalization and stability of bird's nest recognition training network by fine-tuning the trained DSSD network under the condition of insufficient data set. The trained model has faster recognition speed and better robustness, which has important reference significance for the safe and stable operation of OCS.