Design And Realization Of Recognition System Of Museum Visitors’ Violation Behavior

With the development of science and technology and the improvement of people’s living standards,residents are paying more and more attention to the safety of public places.At present,the national government vigorously supports the rapid development of the artificial intelligence industry,and computer vision technology has begun to be applied in various fields of life.How to use deep learning technology to apply to the field of intelligent monitoring has become a key research direction.As an important branch of computer vision,human behavior recognition can not only be developed in medical,autonomous driving and other fields,but also can replace humans to intelligently recognize pedestrian movements in videos,and contribute to improving the safety of national public places.Therefore,how to improve the algorithm performance of human behavior recognition in public places is a research hotspot in the field of computer vision.In order to improve the performance of the human behavior recognition algorithm in the context of public places,this paper proposes a variety of improvements based on the CNN-LSTM network for the contextual feature extraction and training over-fitting of the model.Then,use the improved CNN-BLSTMAD network model to conduct targeted training for the specific scenario of museum visitors’ violations,design and implement an intelligent monitoring system for museum visitors’ violations.Specifically,this article has completed the following work:First of all,in view of the problem of the small human behavior recognition data set,this subject proposes a sliding window data expansion method for the characteristics of human behavior data.The method of frame extraction and frame replacement is used to ensure that the data set is expanded without destroying the data context,thereby improving the effect of model training.Then,we improv the original human behavior recognition model of CNN-LSTM,and the CNN-BLSTMAD network model was proposed.Aiming at the problem of insufficient contextual feature extraction in the model.This topic introduces methods such as deep residual network,two-way mechanism,and deep-level feature extraction to strengthen the model’s ability to extract features.We introduce the Attention mechanism to improve the weight distribution of the areas that should be focused on.Aiming at the over-fitting problem in the model training process,the Dropout mechanism is introduced to improve the training effect of the model.The experimental results show that the improvement strategy used in this subject has improved the recognition accuracy and recognition speed of the original model to a certain extent.Finally,based on the above improved methods,combined with the specific goal of identifying violations of museum visitors,the model is trained and applied on the the scene of tourists’ violation behavior.Then apply the trained model to actual projects,design and implement an intelligent detection system for museum visitors’ violations,and conduct intelligent monitoring and early warning of visitors’ violations.