Pedestrian monitoring method and device

Abstract

The invention provides a pedestrian monitoring method and device. The pedestrian monitoring method comprises the steps of obtaining videos; detecting one or more pedestrians in the videos; identifyingthe identity of each of the one or more pedestrians; determining the motion of at least one video frame having the pedestrian for each of one or more pedestrians; determining whether the pedestrian has abnormal behavior based on the pedestrian identity and the motion of at least one video frame having the pedestrian for each of one or more pedestrians. The invention is advantageous in that whether the pedestrian has abnormal behavior can be determined through comprehensively considering the identity and motion of the pedestrians; the abnormal behaviors can be detected in a more intelligent, highly-efficient and accurate manner; the safety of the area under monitoring can be effectively ensured.

Description

technical field [0001] The present invention relates to the field of computers, and more particularly relates to a pedestrian monitoring method and device. Background technique [0002] In the field of monitoring, it is often necessary to cooperate with a lot of manpower to analyze and maintain the security of the monitored area. Even if many cameras are installed in the monitored area, they are often only used to help the investigation and evidence collection afterwards, and do not really play a role in real-time danger prevention and protection of the safety of the monitored area. With the development of artificial intelligence technology, algorithms and systems have begun to carry out structured processing of video data to realize real-time monitoring of dangerous behaviors. However, in the prior art, the use of video data is often only one-dimensional, such as just doing face detection to determine whether the person entering the monitored area is a suspicious person, o...

AI Technical Summary

Problems solved by technology

However, in the prior art, the use of video data is often only one-dimensional, such as just doing face detection to determine whether the person entering the monitored area is a suspicious person, or just using certain actions of the person entering the monitored area to detect Judging whether a dangerous situation has occurred without actually utilizing the identity information of the person entering the monitored area

Images