High-Altitude Work Safety Belt Usage Action Recognition Dataset

The current security industry faces the challenge of safety hazards for workers performing high-altitude operations, especially regarding the use of safety belts. Existing monitoring methods are mostly static, lacking real-time analysis of worker behavior, and unable to promptly identify and correct safety hazards. This dataset aims to solve this technical problem and business need by recognizing the use of safety belts by workers in high-altitude operations through video data. Data collection involves real-time recording on high-altitude work sites using high-definition cameras, ensuring every important action is fully recorded. Regarding quality control, multiple rounds of annotation and expert review are conducted to ensure data accuracy and consistency. Video data is stored in MP4 format, organized by worker and action. The core advantage of the dataset lies in its high annotation precision and consistency, with annotation accuracy exceeding 95%, and novel data augmentation techniques have enhanced the model's generalization ability. Applying this dataset can effectively reduce high-altitude operation accident rates, increase safety management efficiency, and the ultimate goal is to improve the compliance rate of safety belt usage to 98%.