Serbian Sign Language Dataset

Home » Dataset Download » Serbian Sign Language Dataset

Serbian Sign Language Dataset

Datasets

File

Serbian Sign Language

Use Case

Serbian Sign Language

Description

Explore the Serbian Sign Language Dataset, featuring 33 classes of hand gestures and expressions captured in 2,475 videos.

Description:

The Serbian Sign Language is a comprehensive collection of hand gesture and facial expression videos, meticulously captured to facilitate research and development in gesture recognition and machine learning. This dataset is specifically designed to aid in the creation and training of models that can interpret and understand sign language, providing a valuable resource for developers, researchers, and AI enthusiasts.

Dataset Composition

The dataset consists of 33 distinct classes, each corresponding to a unique hand gesture or facial expression commonly used in Serbian Sign Language. For every class, there are 30 video samples, ensuring diversity and variability in the data. Each video is composed of 75 frames, offering detailed insights into the dynamic nature of sign language gestures. The entire dataset is systematically organized into directories, with each class having its dedicated directory.

Download Dataset

Data Processing and Keypoint Extraction

The videos were capture using a high-definition webcam, ensuring clarity and precision in the record gestures. The raw video data was then processed using the OpenCV library, a robust tool for computer vision tasks, and Mediapipe, a powerful library for real-time hand and face keypoint detection.

The keypoints of the hand in each frame were meticulously extract using Mediapipe, resulting in numpy arrays that map the coordinates of critical hand landmarks. These keypoints are essential for analyzing the hand’s pose, orientation, and movement across frames, allowing for detail gesture interpretation.

Applications

This dataset is a vital asset for various applications, particularly in the fields of gesture recognition, sign language translation, and human-computer interaction. It offers an extensive resource for training machine learning models, enabling advancements in AI-driven sign language recognition systems. Researchers can leverage this dataset to develop algorithms that accurately recognize and translate Serbian Sign Language, contributing to more inclusive communication technologies.