Detecting and Extracting Illegal Signs from Video

Abstract

This project focuses on developing an automated system to detect illegal signs in urban environments from videos. The system utilizes computer vision and machine learning techniques, specifically the YOLOv5 object detection framework, to accurately identify and locate illegal signs in video frames. It incorporates a verification process using Optical Character Recognition (OCR) to differentiate between legal and illegal signs based on the extracted text information. The system is designed as a user-friendly web application, allowing users to upload videos or images for analysis and receive comprehensive results. The system can achieve a detection accuracy of up to 78.6%. With this system, authorities can effectively manage and regulate illegal signs in urban areas, contributing to better urban landscapes.