Recognition About Abnormal Behavior Of Human Based On Video Of Public

With the rapid development of modern video capture technology and computer processing capabilities,the camera can be seen everywhere in life and the Internet has brought the explosive growth of the scale of the video.Faced with the growing number of videos,how to ignore a large number of invalid data in the video,identify the abnormal target from the video quickly and accurately is an important research direction in the field of video recognition.The main research of abnormal behavior recognition in video is analyzing and processing of human behavior in the video sequence data,the characteristics of human behavior extraction in video based on feature recognition of behavior classification,detecting of abnormal behaviors in video timely.The traditional method of abnormal behavior recognition needs to design the action characteristic of the human when the target is extracted from the video.Deep learning algorithm is a very popular research direction in the field of video recognition in recent years.This method does not need the characteristics of artificial design behavior.In this paper,the method of deep learning is used to automatically identify the abnormal behavior of human body in the video.This paper focuses on the design and implementation of the system for identifying the abnormal behavior in the video for the public.The contact theory and practical application,to occur in the complex background of robbery,fighting video more complex behavior of innovative and exploratory research,the abnormal behavior recognition system designed using the improved algorithm;the recognition algorithm is applied in real life.Aiming at the characteristics of abnormal behavior in video,this paper improves the convolutional neural network in the following two aspects.Firstly,the recurrent convolution neural network is used to deal with the correlation between the images in the network,and the LSTM structure is used to solve the problem of the gradient of the recurrent neural network.Secondly,according to the burst of the abnormal behavior in the video,the video is divided into independent units to identify the time of abnormal behavior.Then we test the new convolution neural network on BOSS dataset.The experimental results show that the network has a great progress in the identification performance.In this paper,the algorithm is used in practice,and the new network structure is used to realize the video based public behavior recognition system.The fight,rob,fainting and other behavior are defined as abnormal behavior,the walk,run,wait and other behavior are defined as normal behavior.Analyzing system according to different criteria from structure and process,multiple testing system through the video of the shooting in campus,training and display the identification process of abnormal behaviors in video,finally realized the abnormal behavior recognition function and achieve the goal of research.