sairam333/traffic-congestion-monitoring

🚦 Monitoring Traffic Congestion in Smart Cities Using CNN

<div align="center">

    

</div>

📌 Overview

This project offers a deep learning-based solution for real-time traffic congestion analysis. Using Convolutional Neural Networks (CNNs), the system classifies road images into High Congestion and Low Congestion categories.

We compare a custom CNN against transfer learning models — VGG16, ResNet50, and MobileNetV2 — to find the most accurate architecture.

🌟 Key Features

• —📷 High-Accuracy Classification — Detects congestion states from static images.
• —🔍 Comparative Model Analysis — Evaluates 4 CNN architectures.
• —♻ Transfer Learning — Uses pre-trained models for better performance.
• —🖼 Data Augmentation — Improves robustness and prevents overfitting.
• —🧩 Modular Codebase — Clean scripts for reproducibility.

📊 Results Sneak-Peek

🏆 Best Model: ResNet50 (Transfer Learning) → ~99% Accuracy

[image]

🛠 Project Workflow

1. 1.Data Preprocessing → Organize and augment dataset.
2. 2.Model Training → Train custom CNN and transfer learning models.
3. 3.Evaluation → Generate accuracy reports & confusion matrices.
4. 4.Comparison → Select best-performing model.

📂 Repository Structure

traffic\_congestion\_project/
├── dataset/                # Image dataset
│   ├── High\_Congestion/
│   └── Low\_Congestion/
├── models/                 # Trained .keras models
├── results/                # Evaluation outputs
├── scripts/                # Python scripts
│   ├── preprocess\_data.py
│   ├── train\_cnn.py
│   ├── train\_transfer.py
│   └── evaluate.py
├── README.md
└── requirements.txt

⚙ Getting Started

1️⃣ Dataset Setup

• —Download from Kaggle: Traffic Management - Image Dataset
• —Create folders: dataset/High_Congestion/ and dataset/Low_Congestion/
• —Move:
• —Dense traffic images → High_Congestion
• —Sparse traffic images → Low_Congestion

2️⃣ Installation

# Clone the repo
git clone https://github.com/Sairam-kattunga/Traffic_Congestion_Monitoring_CNN.git
cd Traffic_Congestion_Monitoring_CNN

# Install dependencies
pip install -r requirements.txt

3️⃣ Training Models

cd scripts

# Custom CNN
python train_cnn.py

# Transfer Learning
python train_transfer.py --model vgg16
python train_transfer.py --model resnet50
python train_transfer.py --model mobilenet

4️⃣ Evaluation

python evaluate.py

💡 Future Enhancements

• —🎥 Real-time Video Processing using OpenCV
• —🌐 Web App Deployment (Flask / Streamlit)
• —🚗 Multi-Class Detection (accident, roadwork, fire, etc.)
• —🎯 Hyperparameter Optimization with Optuna / KerasTuner

📜 License

Distributed under the MIT License. See LICENSE for details.

📧 Contact

Rama Venkata Manikanta Sairam Kattunga 🌐 Portfolio 📩 sairamkattunga333@gmail.com 📂 GitHub Repo

>>>>>> 17340f919960658fe9c343dca9540508dbf6d410