电力传输
计算机科学
绝缘体(电)
输电线路
可靠性(半导体)
实时计算
传输(电信)
分水岭
目标检测
人工智能
计算机视觉
功率(物理)
电子工程
模式识别(心理学)
工程类
电气工程
电信
物理
量子力学
作者
Yunhai Song,Gang Zhao,Liang Haoyun,Yanfeng Lu,Sen He,Jianing Shang
标识
DOI:10.1145/3501409.3501608
摘要
Aerial photography patrol has become the main method of electric power patrol on transmission lines instead of manual patrol, and the integrity of insulators on transmission lines directly affects the power supply reliability. Under the interference of complex background, traditional insulator detection method tends to have a low ability to perform the detection tasks. To solve this problem, we proposed a new method based on enhanced YOLOv4. Based on watershed algorithm and data augmentation combinations, the problem of accurate image detection of aerial photo insulator is studied. Combined with the watershed algorithm and data augmentation combinations, the enhanced YOLO model has robust ability against different distances, various lighting conditions in complex environments. It is shown that our proposed method achieved competitive performance in the experiments.
科研通智能强力驱动
Strongly Powered by AbleSci AI