Abstract: This paper proposes a substation defect detection method based on visible and infrared fusion images to improve power inspection efficiency and detection accuracy. The method in this paper uses binocular cameras to acquire visible and infrared images, and employs the Faster R-CNN target detection algorithm to process the fused images, achieving automatic detection of external defects and abnormal temperatures in components such as transformers and insulators. Experimental results show that this method can effectively detect external defects and temperature anomalies.