A monitoring method and system for protecting privacy

Abstract

The present application discloses a monitoring method and system for protecting privacy. The system includes: a monitoring perception module, a feature extraction module, and an abnormal behavior / event detection module; the monitoring perception module is used to obtain monitoring perception data of a monitoring scene; the The feature extraction module is used to extract the real-time feature stream of the monitoring perception data, and sends the real-time feature stream to the abnormal behavior / event detection module; the abnormal behavior / event detection module is used to extract the real-time feature stream Input the pre-trained deep neural network model to detect abnormal behaviors / events. The present invention has the advantages of small amount of feature data and fast transmission, and uses real-time feature stream instead of video, which protects privacy to a certain extent; the camera software of the present invention can be defined, and the model can be updated according to requirements, with wide application range and strong scalability; The feature transformation of the invention can ensure the security of the feature data transmission process, and prevent the cloud from leaking feature data and judgment results; the present invention has cloud computing power.

Description

technical field [0001] The present application relates to the field of computer vision, and in particular to a monitoring method and system for protecting privacy. Background technique [0002] As the most popular direction of computer vision research in recent years, security has a close relationship with video analysis research. In real surveillance video, a common requirement is to automatically identify abnormal events in the video stream, that is, the abnormal event detection task. [0003] Common anomaly detection algorithms learn a common pattern first, and assume that any pattern that violates this common pattern should be abnormal. But in fact, it is difficult and almost impossible for a method to define a so-called normal mode, because the normal mode may contain too many different events and behaviors. Also, it is difficult to define abnormal events, because abnormal events may also include too many types of situations. [0004] Behavior recognition refers to t...

AI Technical Summary

Problems solved by technology

Due to the large amount of audio and video data, it puts a lot of pressure on the cloud server in terms of transmission and storage, and the cloud server itself also has the risk of data leakage, resulting in user privacy leakage

Images