Abstract: Aiming at "unmanned" real-time monitoring and recognition of electrical signs in substations,a simplified YOLOv3 and Tesseract OCR method is proposed. In the light of the particular electrical sign image features,an end point positioning strategy is designed to treat perspective transformation during image pre-processing,and Tesseract OCR engine is used to train the exclusive electrical sign language database,realize the sign detection and recognition task with high accuracy. The experiment results prove the effectiveness of the proposed method.