Patterns Recognition Of Unsafe Behavior In Chemical Laboratory Based On Temporal Segment Network

At present,chemical laboratory accidents at home and abroad account for more than 70%of laboratory safety accidents,and the unsafe behavior in laboratory is one of the main causes of accidents.Therefore,intelligent monitoring and early warning of unsafe behavior of chemical laboratory is implemented.It has important research significance.In this paper,a temporal segment network(TSN)pattern recognition model that fuses attention mechanism and VIT(Vision Transformer)is proposed,and case studies are carried out on public datasets and selfbuilt video datasets of unsafe behavior patterns in chemical laboratories.A pattern recognition system for unsafe behavior in chemical laboratory was designed.The main research contents and results are as follows.(1)A behavior recognition method that fuses attention mechanism and VIT is proposed.Based on the temporal segment network,three attention mechanisms,SENet,ECANet,and CBAM are added to the feature extraction network Res Net respectively,which enhances the spatial feature extraction capability of the feature extraction network,and then integrates VIT and TSN.The spatial feature map is input into VIT to obtain the time series information between the spatial feature maps,which enhances the model’s ability to obtain time series information.(2)Research on pattern recognition of unsafe behavior in chemical laboratory based on temporal segment network fused with attention mechanism and VIT.On the basis of the investigation of unsafe behavior management regulations in chemical laboratories,five unsafe behavior patterns are defined and a video dataset of unsafe behavior patterns in chemical laboratories is constructed.On the dataset of chemical laboratory unsafe behavior patterns,the experiments were carried out using two data division methods,"random division" and "divide according to personnel identity".The results show that the TSN after the fusion of CBAM attention mechanism and VIT achieves 99.0% and 93.2% recognition accuracy in the two groups of experiments,which are 1.0% and 4.3% higher than those before the improvement.The recognition accuracy of the model in both cases is higher than 90%,indicating that the model constructed in this paper can more accurately identify the five chemical laboratory unsafe behavior patterns defined in this paper.(3)A pattern recognition system for unsafe behavior in chemical laboratory is developed based on a temporal segment network fused with attention mechanism and VIT.The system can use the model constructed in this paper to identify unsafe behavior patterns in offline videos,online videos and offline images,providing a concise and effective visualization tool for unsafe behavior pattern recognition for chemical laboratory personnel.The temporal segment network,which integrates attention mechanism and VIT,and unsafe behavior pattern recognition system proposed in this paper,are expected to provide technical support for the monitoring and early warning of unsafe behavior in chemical laboratory...