Retail Theft Detection System Using Yolov8 and Deepsort

Retail Theft Detection System Using Yolov8 and Deepsort

1Cherupally Abhinay, 2Boggula Naga Krishna Reddy, 3Kachu Revanth, 4Kovvuri Divya Sri,

 1,2,3UG Student, 4Assistant Professor, 1,2,3,4Computer Science and Engineering - Artificial Intelligence and Machine Learning, 1,2,3,4Sreenidhi Institute of Science and Technology, Hyderabad, Telangana 501 301.

Abstract

Retail theft, is a major problem faced by businesses across the globe, causing billions of dollars in losses every year. This project proposes an efficient Real-time Retail Theft Detection System that utilizes state-of-the-art computer vision and artificial intelligence (AI) methodologies to actively identify and monitor suspicious behavior in retail settings. By combining the YOLOv8 object detection model with DeepSORT tracking, the system can detect persons and track behavior in video streams. Suspect behaviors, such as frantic hand movements, extended dwell in certain areas, or abnormal path of motion, are automatically triggered. The system comes with a React-based front-end dashboard and a FastAPI back-end for real-time alerting and analytics. Intended for deployment in real-world retail settings, this AI solution is meant to reduce retail loss by allowing for timely intervention.

Keywords: Retail Theft Reduction, YOLOv8, DeepSORT, Object Tracking, Real-time Monitoring, Computer Vision, FastAPI, ReactJS