Multi-region real-time action detection method based on surveillance video

Abstract

The invention discloses a multi-region real-time action detection method based on surveillance video, which has the following steps: a model training stage and a testing stage, wherein, the model training stage is to obtain training data: a marked database of specific actions; The dense optical flow of the video sequence is obtained, the optical flow sequence of the video sequence in the training data is obtained, and the optical flow image in the optical flow sequence is annotated; the target detection model yolo v3 is trained separately by using the video sequence and the optical flow sequence in the training data. , get the RGB yolo v3 model and the optical flow yolo v3 model, respectively. The invention can not only realize the spatiotemporal position detection of a specific action in the monitoring video, but also realize the real-time processing of the monitoring.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to a human motion detection system in a monitoring video scene. Background technique [0002] As the application of monitoring facilities becomes more and more popular, more and more monitoring-based technologies are applied. Action recognition, as one of the most valuable technologies, is mainly used in the interaction of human-machine equipment in indoor and factory environments, as well as in public environments. The security field is used for the detection and identification of specific dangerous actions. [0003] Most of the action recognition methods based on surveillance video mainly focus on the action recognition and classification tasks of the entire scene. Such videos are generally artificially processed video clips, and the video clips generally only contain one type of action, but this kind of video and Natural video clips are very different, and some scholar...

AI Technical Summary

Problems solved by technology

However, the detection of the temporal dimension of actions is still achieved by a multi-scale sliding window on each track, making this method inefficient for longer video sequences

Images