Behavior identification method in dangerous area

Abstract

The invention relates to a behavior identification method in a dangerous area, and the method comprises the following steps: obtaining video data in an actual scene, processing the video data, and arranging the video data into a training sample set which comprises a figure image and position information and category information corresponding to the figure image; training the target detection model by adopting the training sample set to obtain a trained detection model, and then taking the trained detection model as a human body detection model; when the specific position of the image corresponding to the figure is detected, cutting out the image corresponding to the figure to serve as input data of the human body detection model, thereby acquiring key point positions of all parts of the human body; and inferring the behavior characteristics of the figure according to the key point positions. According to the behavior identification method in the dangerous area provided in the technical scheme, the problem that an existing detection method is limited by the conditions of a camera and the image feature extraction detection effect of a field environment can be effectively solved.

Description

technical field [0001] The invention relates to the technical field of target detection, in particular to a behavior recognition method in a dangerous area. Background technique [0002] Call detection is a category of behavior detection, and it is also the direction for the development of artificial intelligence in the future. For the control of public places, such as gas stations, personnel supervision cannot play a real-time supervision role, or post-surveillance video screening requires a lot of Manpower and time cost and not proactive. [0003] The existing call detection technology is basically: 1. The traditional method is to extract the features of the human body, input it into the classifier, and obtain the classification model of whether to make a call; 2. CNN head detection based on deep learning to determine whether there is a call or not. phone behavior. With the vigorous development of deep learning, the behavior detection technology based on deep learning ha...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a behavior recognition method in a dangerous area, which can effectively solve the problem that the existing detection method is limited by the camera's own conditions and the scene environment's detection effect on image feature extraction

Images