🚦 Traffic Sign Classification Using LeNet CNN

A deep learning project implementing LeNet-5 architecture to classify traffic signs from the German Traffic Sign Recognition Benchmark (GTSRB) dataset with 99%+ validation accuracy.

📌 Course Project - Deep Learning & Computer Vision
📁 Repository: ML-project1-Traffic-Sign-Classification

---

📌 Project Highlights

• 🖼️ 43 classes of traffic signs classified
• 🧠 Implemented LeNet-5 CNN architecture from scratch
• 🔍 Achieved 99.2% validation accuracy
• 📊 Comprehensive model performance analysis
• ⚡ Data augmentation techniques applied

---

📂 Dataset Overview

Dataset: German Traffic Sign Recognition Benchmark (GTSRB)

Category	Details
Total Classes	43
Training Images	34,799
Validation Images	4,410
Test Images	12,630
Image Size	32×32 pixels (RGB)

---

🔧 Technical Stack

🧠 Deep Learning

📊 Visualization

🛠️ Utilities

---

🚀 Model Architecture (LeNet-5)

graph LR
    A[Input 32x32x3] --> B[Conv2D 6@28x28]
    B --> C[AvgPool 6@14x14]
    C --> D[Conv2D 16@10x10]
    D --> E[AvgPool 16@5x5]
    E --> F[Flatten 400]
    F --> G[Dense 120]
    G --> H[Dense 84]
    H --> I[Output 43]

Loading

Hyperparameters

• Optimizer: Adam (lr=0.001)
• Loss: Categorical Crossentropy
• Epochs: 50
• Batch Size: 64
• Augmentation: Rotation (±15°), Zoom (±10%)

📊 Performance Metrics

Metric	Training	Validation
Accuracy	99.8%	99.2%
Precision	99.7%	99.1%
Recall	99.6%	99.0%
F1-Score	99.6%	99.0%

Confusion Matrix:

🛠️ Installation & Usage

# Clone repository
git clone https://github.com/mdzaheerjk/ML-project1-Traffic-Sign-Classification.git
cd ML-project1-Traffic-Sign-Classification

Install dependencies

pip install -r requirements.txt

Launch Jupyter Notebook

jupyter notebook "Project 5 - Traffic Sign Classification Using LeNet Network in Keras.ipynb"

📂 Project Structure

ML-project1-Traffic-Sign-Classification/
├── Project/
│   ├── traffic-signs-data/       # Dataset samples
│   ├── Project 5 - Traffic Sign Classification Using LeNet Network in Keras.ipynb
│   └── Project 5 - Traffic Sign Classification Using LeNet.pptx
├── .gitattributes
├── LICENSE
└── README.md

✍️ Author

Mohammed Zaheeruddin

🎓 First-Year B.Tech Student | AI/ML Enthusiast

🏫 Shetty Institute of Technology, Gulbarga

📧 info.zaheerjk@gmail.com

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

text

Key improvements made:
1. Fixed all markdown formatting issues
2. Properly structured tables and code blocks
3. Corrected badge syntax for GitHub/LinkedIn
4. Ensured consistent spacing and headers
5. Maintained all original content while making it properly renderable
6. Fixed the project structure tree formatting
7. Made sure all links and images use proper markdown syntax