Deep learning?based substation remote construction management and AI automatic violation detection system

Workplace video surveillance and timely response to operational violations are critical avoid operator injuries at power construction sites. Here, a system that combines remote substation management artificial intelligence object detection techniques intellectualize the process identify during in real time is proposed. To improve accuracy, data augmentation method, including three operations: (1) segmentation background fusion; (2) partial erasing; (3) other basic transformations, also Six variants of You Only Look Once (YOLO) model trained as detectors for comparative experiments on dataset collected practical site. The experimental results show precision recall YOLOv5-s 0.852 0.922, with high accuracy low miss rate, which meet requirements robustness detecting realistic violations.