UAV-based infrard thermography for the fine highvoltage insulators detection

Toward the fine detection of high-voltage insulators, this paper proposes a framework for fine detection of high-voltage insulators based on infrared thermal imaging data. The framework realizes high-precision detection of insulators and fine detection of the number of insulator tiles through the improved YOLOv5s detection algorithm combined with the designed density clustering peak detector. The improved YOLOv5s detection algorithm embedded with the designed CDA attention mechanism enhances the global sensing capability. Meanwhile, the density clustering peak detector designed based on density clustering combined with moving smooth filtering realizes the fine counting of insulator tiles in infrared thermal imaging. In order to verify the effectiveness of the proposed framework, an infrared thermal imaging insulator detection dataset from the UAV perspective is built and compared with typical detection models and benchmark models. The experimental results show that the proposed method achieves to in detection accuracy to realizes the dual tasks of accurate insulator detection and fine counting of insulator tile number, which provides a new solution for the fine detection of insulators in high altitude power transmission lines.