SATARK (STAY SAFE)

AI assisted distracted driver detection system

Objective: To design and implement a system that uses AI algorithms to detect distracted driving behaviours in real-time.

Key Steps:
Step 1: Data Collection
Step 2: Model Training and Fine-Tuning
Step 3: Real-Time Processing
Step 4: Front-End Development
Step 5: Backend Development
Step 6: Dashboard Development

Takeaways
We were able to train a Resnet neural network model sucessfully by gathering distracted driver dataset.
We used ultra-sonic sensors to detect any nearby objects and give an alert to the driver.
We were sucessful in fetching data from google APIs and recommend nearby hotels for the driver.
We were able to build a database to store numerical aspects of distracted drivers and analyse the data through pie-charts.
We tried exploring explainable AI and assist the drivers.

Block diagram

Installation and Setup

Step 1: Clone the Repository

git clone https://github.com/kramachandrashenoy/Satark.git
cd satark

Step 2: Set Up the Python Environment

1. Create a virtual environment (optional but recommended):

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

2. Install the required Python packages:

   pip install -r requirements.txt

Step 3: Configure MongoDB

1. Set up a MongoDB Atlas account If you don't already have one.
2. Update the MongoDB connection string in the `main.py` file:

   client = MongoClient("your_mongodb_connection_string")

Step 4: Set Up the Ultrasonic Sensors (Arduino)

1. Connect the ultrasonic sensors to your Arduino as per the provided code.
2. Upload the provided Arduino code (`ultrasonic_sensor.ino`) to your Arduino board using the Arduino IDE.
3. Ensure the correct serial port is set in the `main.py` file:

   serial_port = 'COM6'  # For Windows

Step 5: LocationIQ API Configuration

1. Get your API key from LocationIQ.
2. Update the `locationiq_api_key` variable in the `main.py` file with your API key.

Step 6: Run the Flask Application

python main.py

The application will start on `http://127.0.0.1:5000/\`.

Step 7: Access the Application

Open a web browser and go to `http://127.0.0.1:5000/\` to access the main interface.

APIs

Video Feed

• Endpoint: `/video_feed`
• Description: Provides the real-time video feed with drowsiness detection.
• Parameters: `driver_id` (optional, default: 1)

Prediction

• Endpoint: `/prediction`
• Description: Returns the current drowsiness detection status.
• Response: `{ 'prediction': 'Drowsiness Detected' | 'Alert' | 'No Eyes Detected' }`

Ultrasonic Data

• Endpoint: `/ultrasonic`
• Description: Returns the distance measured by the ultrasonic sensors.
• Response: `{ 'distance1': , 'distance2': }`

Current Driving Data

• Endpoint: `/current_driving_data`
• Description: Returns the current driving statistics for the specified driver.
• Parameters: `id` (optional, default: 1)

Total Driving Data

• Endpoint: `/total_driving_data`
• Description: Returns the all-time driving statistics for the specified driver.
• Parameters: `id` (optional, default: 1)

Fetch Nearby Places

• Endpoint: `/fetch_nearby_places`
• Description: Returns the nearby places of interest.
• Response: `{ 'hotel': [...], 'restaurant': [...], 'hospital': [...] }`

Acknowledgements

• Dlib for facial landmark detection.
• Keras for the drowsiness detection model.
• MongoDB for database services.
• LocationIQ for location services.

Contributing

Contributions are welcome! Please open an issue or submit a pull request.