Abnormal behavior recognition based on edge feature and 3D convolutional neural network

With the rapid development of artificial intelligence and 5G technology, the development process of smart city has been greatly accelerated. An important part of the development of smart city is to improve the ability of video monitoring equipment to analyze the content of the video. This paper is to identify the specific abnormal behaviors in the video, including climbing, fighting and falling. The identification of these abnormal behaviors is helpful to find out the danger in time and intervene to ensure personal safety and social stability. Therefore, this paper proposes a deep learning method based on edge extraction. In the first step, the video is decomposed into several frames, and the canny edge detection algorithm is used to extract the edge of the images. In the second step, the extracted edge information is used as the input of the 3D convolutional neural network, and the 3D convolutional neural network model is trained iteratively to recognize the abnormal behaviors. Through a large number of experiments, the effectiveness of the proposed algorithm is verified.