Research On Detection Of Abnormal Behavior In Classroom Monitoring Video

Nowadays surveillance video has been widely applied in daily life,and the function of intelligent video surveillance system is far from meeting people's needs: real-time and accurate target detection,precise abnormal behavior warning,complete behavior statistics and so on.The development of intelligent monitoring video system has been paid attention by many researchers,and the development of related technology still has great improvement space.In the aspect of behavior recognition,related research has paid attention to convolution neural network algorithm because of its intelligent feature extraction method and good recognition ability in complex scenes.In this paper,video monitoring abnormal behavior detection project,designed according to the functional requirements of the system,with deep learning is proposed based on the neural network algorithm for real-time detection.Moreover,the method of common recognition of character behavior and interactive object is put forward,and it is verified by experiment.The main research work is as follows:(1)For classroom monitoring video abnormal behavior detection scene,the deep learning convolutional neural network structure achieves a better behavior recognition effect than traditional algorithms.The traditional convolution neural network classification algorithm is replaced by regression algorithm,which improves the speed of target detection,and meets the requirements of convolution neural network's behavior detection speed to the real-time behavior of video surveillance.(2)The relationship between the behavior of a person in a real scene and the object around the person is considered into the neural network algorithm.The behavior features and the interaction object feature extracted from the neural network in the scene using SPP technology combine features with different parameters.What's more,calculate the conbined features of the characteristics of the behavior and the characteristics of the interaction.(3)According to the monitoring video abnormal behavior detection algorithm proposed in this paper the existing difficult problems of small object recognition,mask module is introduced in the image processing algorithm.Using the behavior feature and interactive object feature combination algorithm,further improve the recognition rate of human action.Through the theoretical analysis and experiment,this paper proposes the semantic behavior identify convolution neural network is suitable for the classroom monitor video abnormal behavior detection,has high robustness,identify the characteristics of fast speed and high recognition rate,on the depth of similar complex scene monitoring system of further research of learning algorithm has important significance.