🚦 Traffic Sign Recognition using Deep Learning (CNN)

This project implements a Traffic Sign Recognition System using Convolutional Neural Networks (CNNs) and Computer Vision techniques. It is capable of classifying traffic signs from images and also supports real-time recognition using a webcam.

The project demonstrates a complete deep learning workflow: dataset handling → model training → evaluation → real-time inference.

📌 Features

Image classification using CNN (TensorFlow / Keras) Preprocessing and normalization of input images Performance evaluation using accuracy and confusion matrix Real-time traffic sign recognition via webcam Confidence thresholding and temporal smoothing for stable predictions

🛠 Tech Stack

Python TensorFlow / Keras OpenCV NumPy Matplotlib / Seaborn

📂 Project Structure

TrafficSignRecognition/ │ ├── dataset/ │ └── README.md # Dataset instructions (full dataset not included) │ ├── Traffic_sign_Reco.py # Model training and evaluation ├── webcam_predict.py # Real-time webcam prediction ├── check_structure.py # Dataset structure validation ├── requirements.txt # Project dependencies ├── .gitignore └── README.md

📊 Dataset

This project uses a custom traffic sign image dataset. ⚠️ The full dataset is not included in the repository due to size constraints. You can generate or collect your own images following the folder structure described below.

Expected Dataset Structure dataset/ └── raw_images/ ├── Stop/ ├── Speed_Limit_50/ ├── Yield/

🚀 How to Run the Project

1️⃣ Install Dependencies pip install -r requirements.txt

2️⃣ Train the Model python Traffic_sign_Reco.py

This will: Train the CNN model Save the trained model as final_traffic_sign_model.h5

3️⃣ Real-Time Traffic Sign Recognition (Webcam) python webcam_predict.py

Controls inside webcam window: q → Quit s → Save current frame r → Reset predictions t → Toggle confidence threshold d → Debug information

📈 Results

High classification accuracy on validation data Stable predictions using confidence thresholding Smooth real-time recognition using temporal averaging (Sample result images and confusion matrix can be added in a results/ folder.)

🎯 Learning Outcomes

Practical implementation of CNNs for image classification Dataset preprocessing and augmentation techniques Model evaluation and performance visualization Real-time deployment of deep learning models using OpenCV

📌 Future Improvements

Data augmentation for better generalization Transfer learning with pretrained models (ResNet, MobileNet) Deployment as a web or mobile application