metadata
pretty_name: SignLanguageKeypoints
task_categories:
- video-classification
tags:
- hands
- sign-language
- landmarks
Dataset used for a training a model to classify Danish Sign Language signs, based on MediaPipe hand landmark data.
The data is not split into training, test and validation sets.
The dataset consist of four classes, 'unknown', 'hello', 'bye' and 'thanks'.
There are 30 datapoints for each class.
Each data point is 30 frames of data stored in individual Numpy files with x, y and z values for each hand landmark.