Fatigue / Driver Drowsiness Prediction

This repository collects data and experiments for detecting driver fatigue and drowsiness (yawning / eye-closure) using CNN models. The data and baseline model/code referenced here are based on the Kaggle notebook: https://www.kaggle.com/code/yaswanthkumarmalli/cnn-model/notebook

Credit

• The datasets and initial model/code were obtained from the Kaggle notebook by yaswanthkumarmalli. If you use this work, please cite and link to that Kaggle page.

Overview

• Goal: train a convolutional neural network (CNN) to detect/predict driver drowsiness/yawning from images.
• This workspace contains datasets gathered from multiple public sources and an experimental notebook implementing a CNN.

Repository layout

• data/ - raw datasets used for experiments. Subfolders include:
  • Detect-Yawning/ (train/test splits, CSVs)
  • Driver Drowsiness Dataset (DDD)/ (Drowsy / Non Drowsy)
  • nthuddd2/ (drowsy / notdrowsy)
  • Yawn Dataset/ (yawn / no yawn)
• models/ - saved model checkpoints (if present)
• scripts/ - scripts and notebooks used for training and evaluation
  • cnn-model-v8.ipynb - main notebook used to train/evaluate the CNN model

Quick start

1. Install dependencies listed in pyproject.toml (or set up a Python 3.8+ venv and install required packages such as numpy, pandas, tensorflow/torch, and jupyter).
2. Open the notebook scripts/cnn-model-v8.ipynb in Jupyter or VS Code to reproduce training and evaluation steps.
3. Ensure the data/ folder contains the dataset directories (already present in this repo). Adjust any dataset paths inside the notebook if needed.

License

• Check the original dataset and Kaggle notebook licenses before redistribution. This repository itself does not include a license file; add one if you intend to release the code publicly.

Acknowledgements

• Thanks to the contributors of the datasets and to the Kaggle notebook author (yaswanthkumarmalli) for the baseline CNN implementation used as reference.