drowsiness-validator 0.1.4

Drowsiness detection using facial landmarks or CNN.

Drowsiness Validator

A Python package for detecting drowsiness using facial landmarks or a Convolutional Neural Network (CNN).

Installation

Install the package using pip:

pip install drowsiness-validator

Usage

Here's a basic example of how to use the package:

from drowsiness_validator import detect_drowsiness

# Example with image path
test_img_path = 'path/to/your/image.jpg'
result = detect_drowsiness(image_path=test_img_path, method='cnn')
print(result)
# Example output: {'prediction': 1, 'status': 'Drowsy', 'method': 'cnn'}

# Example with base64 image
import base64
with open(test_img_path, 'rb') as f:
    img_b64 = base64.b64encode(f.read()).decode('utf-8')
result = detect_drowsiness(image_base64=img_b64, method='cnn')
print(result)
# Example output: {'prediction': 1, 'status': 'Drowsy', 'method': 'cnn'}

Flask Route Example

Here is an example of a Flask route to check drowsiness from a base64 image:

from flask import Flask, request, jsonify
from drowsiness_validator import detect_drowsiness

app = Flask(__name__)

@app.route('/check-drowsiness', methods=['POST'])
def check_drowsiness():
    data = request.json
    if 'image_base64' not in data:
        return jsonify({'error': 'image_base64 is required'}), 400

    image_base64 = data['image_base64']
    result = detect_drowsiness(image_base64=image_base64, method='cnn')
    return jsonify(result)

if __name__ == '__main__':
    app.run(debug=True)

Documentation

detect_drowsiness

Detects drowsiness from an image using either CNN or facial aspect ratios (via Random Forest).

Args:

• image_path (str, optional): Path to the input image file. Defaults to None.
• image_base64 (str, optional): Base64 encoded string of the input image. Defaults to None.
• method (str, optional): Method for detection ('cnn' or 'aspect_ratio'). Defaults to 'cnn'.
• force_train_cnn (bool, optional): Whether to force retraining of the CNN model. Defaults to False.
• train_dir (str, optional): Path to the training data directory (needed if force_train_cnn=True).
• test_dir (str, optional): Path to the test data directory (needed if force_train_cnn=True).

Returns:

• dict: A dictionary containing the prediction result and method used.

Example Response:

{
  "prediction": 1,
  "status": "Drowsy",
  "method": "cnn"
}

calculate_ARs

Calculates aspect ratios from an image.

Args:

• image_path (str, optional): Path to the input image file. Defaults to None.
• image_data (numpy array, optional): Image data. Defaults to None.

Returns:

• dict: A dictionary containing calculated aspect ratios or an error message.

  Example:

  {
    "EAR": 0.289,  # Eye Aspect Ratio
    "MAR": 0.625,  # Mouth Aspect Ratio
    "MOE": 2.162,  # Mouth over Eye Ratio
    "HPR": 1.342,  # Head Pose Ratio
    "BAR": 12.456  # Brow Aspect Ratio
  }
  

calculate_all_ARs

Calculates all aspect ratios from either an image path or image data (numpy array).

Args:

• image_path (str, optional): Path to the input image file. Defaults to None.
• image_data (numpy array, optional): Image data. Defaults to None.

Returns:

• dict: A dictionary containing calculated aspect ratios.

predict_drowsiness_with_cnn

Predicts drowsiness using the trained CNN model from image data or path.

Args:

• image_data (numpy array, optional): Image data as numpy array. Defaults to None.
• image_path (str, optional): Path to the input image file. Defaults to None.
• force_train (bool, optional): Whether to force retraining of the CNN model. Defaults to False.
• train_dir (str, optional): Path to the training data directory (needed if force_train=True).
• test_dir (str, optional): Path to the test data directory (needed if force_train=True).

Returns:

• tuple: (prediction, confidence) where prediction is 0 (Active) or 1 (Drowsy), and confidence is a float between 0 and 1.

Author

Sharjeel Baig

• Portfolio: https://sharjeelbaig.github.io
• GitHub: https://github.com/sharjeelbaig

License

This project is licensed under the MIT License.