An electric bicycle abnormal behavior detection method and system

CN115171056BActive Publication Date: 2026-07-03ZHEJIANG SOS TECH

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Patents(China)
Current Assignee / Owner: ZHEJIANG SOS TECH
Filing Date: 2022-07-06
Publication Date: 2026-07-03

AI Technical Summary

Technical Problem

Existing technologies lack effective methods to predict abnormal riding behavior of electric bicycles under nighttime conditions. In particular, the characteristics of electric bicycles are difficult to analyze, and their riding routes do not have obvious purpose, making them difficult to identify in video surveillance.

Method used

By acquiring and preprocessing surveillance video data, the VBM3D algorithm is used to remove noise and histogram equalization is used to enhance video quality. The ResNet50 and SiameseRPN++ algorithms are combined to identify electric bicycles. The license plate color is outlined using cascade faster RCNN. The MapReduce algorithm is used to aggregate the data. Suspicious electric bicycles are identified based on riding time and license plate color, and abnormal behavior is determined by speed differences.

Benefits of technology

It enables rapid prediction of abnormal behavior of electric bicycles, and can identify and confirm the abnormal behavior of suspicious electric bicycles and their accompanying vehicles under nighttime conditions, thus improving the accuracy and efficiency of electric bicycle behavior analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure CN115171056B_ABST

Patent Text Reader

Abstract

This invention provides a method for detecting abnormal behavior of electric bicycles, including acquiring and preprocessing surveillance video data, and extracting video data showing at least two vehicles simultaneously loitering in a designated area from the preprocessed surveillance video data; if the analysis reveals that at least two vehicles are electric bicycles, determining the riding loitering time and corresponding license plate color of each electric bicycle, and identifying electric bicycles whose riding loitering time and license plate color meet predetermined combination conditions as suspicious electric bicycles; determining the speed of each suspicious electric bicycle and the speeds of each subsequent electric bicycle, and when the speed difference between a suspicious electric bicycle and at least one subsequent electric bicycle is within a predetermined value range, identifying the suspicious electric bicycle with a speed difference within the predetermined value range as an electric bicycle with abnormal behavior. Implementing this invention can predict abnormal electric bicycle riding behavior in advance.

Need to check novelty before this filing date? Find Prior Art