Abnormal behavior detection method based on deep convolutional neural network

Abstract

The invention discloses an abnormal behavior detection method based on a deep convolutional neural network. The method comprises the steps of A1, encoding an input video frame; A2, decoding the codedstream to obtain an appearance stream and a motion stream; and A3, scoring the frames through an anomaly detection module, comparing the scores with a threshold value, and judging abnormal behaviors.According to the method, the structure information and the motion information extracted from the video frame are fully utilized, and intelligent detection of abnormal behaviors can be accurately and efficiently completed.

Description

technical field [0001] The invention relates to the fields of computer vision and video detection and analysis, in particular to a method for detecting abnormal behavior based on a deep convolutional neural network. Background technique [0002] The goal of a practical anomaly monitoring system is to be able to send a signal in time once an abnormal situation occurs, and to identify the category of the anomaly. Overall, anomaly detection can be seen as rough video understanding, which just distinguishes abnormal from normal. Once abnormalities are detected, further classification techniques are used to identify and classify abnormal behaviors. [0003] In order to realize the online detection of abnormal behavior in video surveillance, the following three difficulties need to be overcome: the algorithm can meet the real-time requirements; the algorithm can effectively use long-sequence uncropped video data sets; the algorithm can cope with the complexity of the environment ...

AI Technical Summary

Problems solved by technology

However, this approach is inefficient, and estimating optical flow is often computationally and storage-costly

Images