redeagle17/drowsiness-detection

Drowsy Face Detection and Classification with YOLO

Overview

This project aims to enhance safety in critical environments by developing a computer vision system that can detect and classify human faces as either drowsy or awake. It utilizes the YOLO (You Only Look Once) object detection model and a custom dataset generated by capturing images from a webcam and categorizing them with different labels.

Table of Contents

• Getting Started
• Prerequisites
• Dataset Collection
• Training

Getting Started

Follow these instructions to set up and run the project on your local machine.

Prerequisites

• Python 3.7
• OpenCV
• YOLO (You will need the YOLO weights and configuration files)
• Webcam (for capturing images)

Install OpenCV using pip:

pip install opencv-python

Dataset Collection

To train a drowsy-awake classifier, a custom dataset needs to be collected. Follow these steps to create your dataset:

• Set up a webcam or camera.
• Capture images of faces in various states (drowsy and awake).
• Organize the images into appropriate folders, e.g., dataset/drowsy and dataset/awake.

Training

Train the YOLO model with your custom dataset:

• Download the YOLO weights and configuration files.
• Configure YOLO for training using the provided configuration file.
• Start the training process by running the YOLO training script with your custom dataset

!yolo task=detect mode=train model=yolov5s.pt data=../content/drive/MyDrive/yolov5CustomTraining/dataset.yaml epochs=50 imgsz=320

Usage

Once the YOLO model is trained, you can use it for real-time drowsy-awake face detection and classification. Run the following command:

python main.py