Real-Time drowsiness Detection with YOLOv5 and PyTorch

Identification of objects in an image considered a common assignment for the human brain, though not so trivial for a machine. Identification and localization of objects in photos is a computer vision task called ‘object detection’. One of the most popular algorithms to date for real-time object detection is YOLO (You Only Look Once).

In this project, we performed drowsiness detection to check whether a person is awake or drowsy, using the latest YOLOv5 implementation developed by Ultralytics.

Prerequisites

• PyTorch
• Torchvision
• Torchaudio
• OpenCV
• Matplotlib
• Numpy
• ipywidgets
• PyQt5
• lxml

Installation

1. Install the required dependencies by running the following command:

   !pip3 install torch torchvision torchaudio

   !pip install opencv-python matplotlib numpy ipywidgets

   !pip install pyqt5==5.15.2 lxml

2. Clone the YOLOv5 repository by running:

   !git clone https://github.com/ultralytics/yolov5

3. Navigate to the cloned directory:

   cd yolov5

4. Install the requirements for YOLOv5:

   !pip install -r requirements.txt

5. Clone the labelImg repository for image labeling:

   !git clone https://github.com/tzutalin/labelImg.git

Usage

1. Collect Images:

• Connect a webcam to your computer.
• Run the code to capture images for the specified labels ("awake" and "drowsy").
• Images will be saved in the "data/images" directory.

2. Label Images:

• Open the labelImg tool to label the collected images as "awake" or "drowsy".
• Save the labeled annotations as XML files.

3. Train the YOLOv5 Model:

• Run the training script with the specified parameters (image size, batch size, epochs, data configuration, weights, etc.).
• The model will be trained on the labeled images using the YOLOv5 architecture.

4. Load the Trained Model:

• Load the trained model weights for inference.
• Perform object detection on images or video streams.