HOI-based Method For Detecting Cell Phone Behavior In Industrial Safety Production Environments

With the increasing popularity of mobile payment and other technologies,smartphones have become an indispensable part of people’s daily lives in China,which is also the world’s largest industrial manufacturing country.However,as China’s industrial production capacity continues to rise,safety issues in the industrial sector have become increasingly prominent.Violations of cell phone use in industrial production environments have led to safety accidents and even casualties.Therefore,it is of great practical significance to study human-object interaction detection algorithms that can achieve fast and accurate detection of such violations,thus enhancing safety supervision.To address the practical problems of algorithm inference timeliness and the complex environment in industrial settings,this thesis proposes an end-to-end human-object interaction detection algorithm based on Transformer’s attention mechanism.By sampling key features at different scales from the industrial complex environment and using Transformer’s attention mechanism to aggregate global information,the algorithm achieves accurate detection of targets in complex environments while eliminating post-processing processes such as non-maximal value suppression to achieve real-time detection.Moreover,a deformable attention module is used to optimize small target detection and the multi-headed attention mechanism of Transformer,which adaptively adjusts the position and shape of the feature extraction window,reducing the computational effort of the attention computation process and accelerating the model training time.Experimental results demonstrate that the proposed HOI-based method for detecting cell phone violations in industrial safety production environments has higher detection accuracy and significant advantages in inference speed compared with existing advanced algorithms on the HICO-DET dataset after expanding the cell phone category.The proposed algorithm is also deployed in a web application with front and back ends,which allows users to easily perform HOI detection of cell phone violations in industrial environments through web pages.