jasir/palsynet-data

Dataset Card for Dataset Name

A data set of images of faces of people affected with Bell's palsy (Facial palsy).

Dataset Details

Dataset Description

A data set of images of faces of people affected with Bell's palsy (Facial palsy). Created using curating and editing publically available youtube videos. Also included are images from people not affected by it, using the same method.

• License: CC-BY-4.0

Uses

Can be used to train image models to detect and classify signs or intensity of bell's palsy on facial images.

Dataset Structure

The data is provided as videos, one per person. A script is provided to extract individual frames from those videos to create an image dataset for further processing and training.

To extract the images, you can use the provided script extract_images.py. The provided requirements.txt file can be used to install the required dependencies for the script. (namely opencv-python and rich)

You can configure the output directory and the number of frames to skip b/w consecutive images in the script.

pip install -r requirements.txt
python extract_images.py