👁️ Computer Vision

Vehicle Detection & Tracking

Multi-Class Vehicle Tracking for Traffic Analysis

2 months Solo Project Completed

Overview

Real-time detection and tracking system distinguishing between cars, buses, motorcycles, and trucks for traffic flow optimization and analytics.

The Problem

Traffic management requires accurate counts of vehicles by type to optimize signal timing and understand traffic patterns.

The Solution

Implemented YOLOv8 + DeepSORT pipeline with custom vehicle classification. System counts vehicles crossing virtual lines and generates traffic density heatmaps.

Results & Impact

92% Counting Accuracy Across all vehicle types
25 FPS Processing Speed On edge device
4 Vehicle Classes Car, bus, bike, truck

Key Impact

  • Data used for city traffic flow optimization
  • Enabled data-driven signal timing adjustments
  • Reduced congestion by 18% at monitored intersections

Technologies Used

PythonYOLOv8DeepSORTOpenCVNumPy