A method for detecting human body abnormal behaviors in an indoor scene

Abstract

The invention discloses a method for detecting human body abnormal behaviors in an indoor scene, and relates to the technical field of human body abnormal behavior detection. According to the method, a multi-target tracking algorithm FairMOT, a human body posture recognition algorithm AlphaPose and an SVM (Support Vector Machine) model are combined, human body feature information is extracted by using joint point coordinate data, and human body actions are marked, so that abnormal actions can be effectively recognized in an indoor environment, and the detection speed and precision are improved while environmental factor interference is avoided.

Description

technical field [0001] The invention relates to the technical field of abnormal human behavior detection, in particular to a method for detecting abnormal human behavior in an indoor scene. Background technique [0002] With the gradual development of modern social science and technology, monitoring equipment such as cameras has gradually played a vital role in many places, and analyzing abnormal human behavior from video monitoring has become a current research hotspot. [0003] However, the existing human abnormal behavior detection methods generally process and classify the human body feature information extracted directly from the image in the early stage. For example: by real-time tracking and detection of the human body's motion trajectory and setting a threshold to determine whether the motion trajectory is a normal trajectory or an abnormal trajectory, in order to detect abnormal behaviors, because the set threshold is difficult to distinguish between normal and abno...

AI Technical Summary

Problems solved by technology

For example: by real-time tracking and detection of the human body's motion trajectory and setting a threshold to determine whether the motion trajectory is a normal trajectory or an abnormal trajectory, in order to detect abnormal behaviors, because the set threshold is difficult to distinguish between normal and abnormal trajectory, resulting in its The detection ability is limited and the accuracy is not high; the existing methods of automatic detection of abnormal behavior based on video processing are greatly affected by the video environment during the detection process, and a variety of features need to be extracted during the detection process, resulting in a slow detection speed; The multi-feature human distraction behavior detection algorithm based on convolutional neural network is only for single-person anomaly detection. When there are too many people and the environment is complex, the detection accuracy will also drop greatly.

In addition, most researchers currently use image information as the direct research object, which will lead to the extraction of human body feature information subject to the influence of different appearances, viewing angles, occlusions, etc. in the image, which will bring great difficulties to the analysis process

Images