Back to Summary
// Final Year Project

Smart Traffic Violation Detection

Real-time multi-camera monitoring system processing 4 concurrent streams at ~30 FPS for smart city infrastructure.

01.The Problem

Traditional traffic monitoring systems struggle with high latency and inaccurate speed tracking across multiple concurrent video streams in real-time.

02.The Solution

Built a multi-camera system processing 4 streams at ~30 FPS using YOLOv8 and ByteTrack with dynamic frame skipping. Developed a speed detection engine using pixel-to-meter scaling geometry and exponential smoothing filters. Optimized latency with an asynchronous, multi-threaded ALPR pipeline utilizing PaddleOCR, all managed through a Flask/SQLite central administrative dashboard.

System Architecture

System Data

RoleAI Engineer
StatusIn Development
Tech Stack
YOLOv8ByteTrackPaddleOCRFlaskSQLiteChart.js