TrafficVisionAI

This project implements a real-time driver assistance system using traffic sign recognition, lane detection, and collision detection.

Directory Structure

TrafficVisionAI/
├── data/
│   ├── GTSRB/
│   │   ├── train/
│   │   │   ├── 0/
│   │   │   ├── 1/
│   │   │   └── ...
│   │   ├── Meta.csv
│   │   ├── Test.csv
│   │   └── Train.csv
├── models/
│   └── traffic_sign_model.h5
├── src/
│   ├── __init__.py
│   ├── traffic_sign_recognition.py
│   ├── lane_detection.py
│   ├── collision_detection.py
│   ├── main.py
│   ├── train_model.py
├── utils/
│   ├── __init__.py
│   ├── preprocess_data.py
│   ├── helpers.py
│   └── config.py
├── .gitignore
├── README.md
└── requirements.txt

Setup

1. Clone the Repository

git clone https://github.com/debjit-mandal/TrafficVisionAI.git
cd TrafficVisionAI

2. Create a Virtual Environment and Activate It

python -m venv env
source env/bin/activate  # On Windows use `env\Scripts\activate`

3. Install the Required Packages

pip install -r requirements.txt

4. Download and Extract the GTSRB Dataset

Download the GTSRB dataset and extract it into the data/GTSRB directory.

5. Train the Traffic Sign Recognition Model

Train the model using the GTSRB dataset:

python -m src.train_model

6. Run the Real-Time Driver Assistance System

Start the system to recognize traffic signs, detect lanes, and perform collision detection:

python -m src.main

Usage

The system will start the webcam feed, process each frame to recognize traffic signs, detect lanes, and perform collision detection. Press q to quit.

Project Components

1. Traffic Sign Recognition The traffic sign recognition component uses a convolutional neural network (CNN) trained on the GTSRB dataset. The trained model is used to recognize traffic signs in real-time from a webcam feed.

2. Lane Detection The lane detection component uses OpenCV to detect lane lines in real-time from a webcam feed. It applies Canny edge detection and Hough line transform to identify lane lines.

3. Collision Detection The collision detection component uses a pre-trained object detection model from TensorFlow Hub to detect objects in real-time from a webcam feed. It draws bounding boxes around detected objects and calculates distances.

Dependencies

• TensorFlow
• TensorFlow Hub
• OpenCV
• NumPy
• Pandas
• Scikit-learn

Files and Scripts

1. src/train_model.py Script to train the traffic sign recognition model using the GTSRB dataset.

2. src/traffic_sign_recognition.py Script to recognize traffic signs in real-time using the trained model.

3. src/lane_detection.py Script to detect lane lines in real-time using OpenCV.

4. src/collision_detection.py Script to detect objects in real-time using a pre-trained model from TensorFlow Hub.

5. src/main.py Main script to integrate all functionalities: traffic sign recognition, lane detection, and collision detection.

6. utils/preprocess_data.py Utility script for preprocessing images and loading the GTSRB dataset.

7. utils/helpers.py Utility script containing helper functions for lane detection.

8. utils/config.py Configuration file containing paths and settings.

Contributing

Contributions are welcome! Please fork this repository and submit pull requests.

License

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

---