A Drowsiness Driving Alert implementation with the use of MediaPipe
Explore the docs »
View Demo
·
Installation
·
Contact
Table of Contents
The following repository describes a computer vision project with the objective of driver drowsiness detection by implementation of CNNs (Convolutional Neural Networks), developed by students of the Polytechnic University in Madrid (UPM) as part of the course 'Computer Vision'. In this repo it can be found the libraries, results and a collection of the reports produced during the development of this project.
List of major frameworks/libraries used for this project.
In order to reproduce the Drowsiness System Detection you can clone this repo:
git clone https://github.com/jbarciv/DrowsyDrive-Alert.git
or just download the Notebook for the alarm system along with the MediaPipe face landmarker.
All the system has been develop on Ubuntu 20.04. Although it has not been tested, the following process should be practically the same for Windows except for minor details. We do not assure compatibility with Windows but we believe it is easily achievable.
We recommend to create a virtual environment in which to run the Notebook.
- Create the venv:
python3 -m venv myvenv
- Activate your venv:
source myvenv/bin/activate
we recommend you to create and alias in your .bashrc file:
alias myvenv='source ~/the_path_to_venv_folder/myvenv/bin/activate'
Only the Jupyter Notebook and some basic libraries are needed, you can install them easily in your venv using pip:
- Install Jupyter Notebook:
pip install notebook
- Install the libraries:
pip install opencv-python mediapipe
With the use of MediaPipe we have implemented a driving drowsiness detection system. It relies only on visual perception from a camera (potencially the use of the front phone camera). The systems expects to have a close vision of the driver and a daylight conditions. The systems has not been tested with IR cameras for a robust light conditions.
Distributed under the BSD 3-Clause License. See LICENSE.txt for more information.
Ivonne Dayanna Quishpe Villagomez - [email protected]
Juan Gustavo Maldonado Quispe - [email protected]
Micaela Cabrera Guerrero - [email protected]
Jorge Guijarro Tolon - JTlotus - [email protected]
Josep Mª Barberá Civera - jbarciv - [email protected]
Visit our project website for more info: https://driver-drowsiness-detect.webflow.io/
Here we list resources we find helpful and would like to give credit to.