Stranger intrusion detection method based on cloud-edge collaboration

Abstract

The invention provides a stranger intrusion detection method based on cloud-edge collaboration, which realizes the detection, positioning and tracking functions of non-park staff. It includes the following steps: establishing a face database of park staff; deploying a face detection and face recognition model on the terminal at the entrance of the park to detect and mark non-park staff; terminal data collection and preprocessing, and uploading to the edge server ;The edge server screens the target images from multiple video sources, and records the frame image time and motion trajectory information; integrates target image sequences of different source videos, generates their motion trajectories, and feeds back the results; analyzes and processes the processing results regularly, Feedback on the processing results whose recognition accuracy is lower than the threshold, upload the error data to the cloud, update and train the model in the cloud, and feed back the training parameters to the edge server for update.

Description

technical field [0001] The invention relates to a stranger intrusion detection method based on cloud-edge collaboration, and belongs to the technical field of face recognition. Background technique [0002] Park security upgrade, how to use artificial intelligence technology to realize the monitoring and early warning of the park is a problem that needs to be solved urgently, mainly including identity recognition, stranger intrusion warning, cross-camera recognition and trajectory tracking and other functions. [0003] The main technical difficulty lies in the realization of cross-camera target positioning and real-time analysis combined with pedestrian re-identification technology. Due to the location, angle and other reasons, the cameras in the park cannot always capture the face information of pedestrians. Therefore, it is usually unrealistic to rely on face recognition to find the same pedestrian in multiple cameras. Clothes will not be changed in a short period of time...

AI Technical Summary

Problems solved by technology

[0004] Finally, huge data storage restricts the development of intelligent computing. Cameras are installed more and more, and devices acquire more data than before, but not all data is valuable. Therefore, it is particularly important to decide how to integrate and store data more effectively.

Images