🚗 SmartFlow is a computer vision-based intelligent traffic management system that detects, classifies, and analyzes vehicle flow in real time to optimize traffic efficiency and reduce congestion.
SmartFlow leverages Deep Learning (YOLOv8) and Computer Vision (OpenCV) to detect vehicles from real-time traffic footage.
The system computes lane-wise density, transmits live data via APIs, and is designed to power adaptive traffic signals in the future.
📸 Dataset Summary:
- Total images: 8105
- 📂 6484 for training
- 📁 1621 for validation/testing
🎯 Classes Detected:
🚗 Cars 🚌 Buses 🏍️ Motorbikes 🚛 Trucks 🚐 Vans 🚲 Bicycles 🛺 Rickshaws
💻 Technical Details
| Parameter | Value |
|---|---|
| Framework | Ultralytics YOLOv8L |
| GPU Used | NVIDIA GeForce RTX 3050 (6 GB) |
| Training Duration | ~2 Days |
| mAP50 (Cars) | 0.753 |
| Overall mAP50–95 | 0.576 |
| Inference Speed | ⚡ 0.3 ms preprocess | 40.4 ms per image |
Developed a real-time detection and analytics system using:
- 🧩 OpenCV + Python Multithreading for parallel frame processing
- 🌐 Backend API Integration for seamless data flow
- 📊 Optimized visualization for lane density & traffic metrics
The result is low-latency, high-speed detection ideal for smart city applications.
flowchart LR
A[📷 Traffic Camera Feed] --> B[🧠 YOLOv8 Detection]
B --> C[📈 Lane-wise Vehicle Count]
C --> D[🌐 API Data Transmission]
D --> E[📊 Dashboard Visualization]
E --> F[🚦 Adaptive Signal Optimization]