Ziqiang Zhang
Object detection, safety helmet detection, personnel clothinginspection, personnel detection, bill detection
Machine learning is playing an increasingly important role in smart substation systems. Object detection algorithms are commonly used in smart substations for procedures, such as helmet detection and personnel clothing inspection. However, object detection algorithms are inadequate for solving complex smart substation scenarios because of their poor generalisation ability. Thus, we introduce an intelligent fusion algorithm named YYSF-4 that has good generalisation ability. YYSF-4 comprises You Only Look Once (YOLO) V1, YOLO V3, a single-shot multi-box detector, and fast- oriented text spotting, and is suitable for use in smart substations. We use real images from substations as a dataset to verify the effectiveness of the YYSF-4 in four scenarios: helmet detection and recognition, personnel clothing detection and identification, personnel detection and identification, and bill detection and recognition. The experimental results show that the mean average precision (mAP) of YYSF-4 in the above four scenarios is higher than the mAPs of other baseline algorithms.
Important Links:
Go Back
