Signclusive Mediapipe

Home » Dataset Download » Signclusive Mediapipe

Signclusive Mediapipe

Datasets

File

Signclusive Mediapipe

Use Case

Signclusive Mediapipe

Description

Discover the Signclusive Mediapipe dataset: 13,500 images of hand signs for all letters and the space sign, collected from five diverse signers.

Description:

The Signclusive Mediapipe dataset is a comprehensive collection designed for the development and training of machine learning models in recognizing sign language. This dataset encompasses images representing the 26 letters of the English alphabet, as well as the “space” sign, making a total of 27 distinct classes. Each class is represented by images from five different signers, ensuring diversity in hand shapes and styles.

Download Dataset

The data collection process for the Signclusive Mediapipe dataset involved gathering images from five different signers, each contributing 100 samples per letter of the alphabet and the “space” sign, resulting in 500 samples per letter and a total of 13,500 images. Each image was originally captured at a resolution of 640×480 pixels and later resized to 224×224 pixels to meet the input requirements for neural network training. This resizing ensured consistency and compatibility across the dataset. Additionally, the images were processed using the Mediapipe framework to detect and highlight hand landmarks, with the background blacked out to focus the model’s attention solely on the hand features, enhancing the model’s ability to accurately recognize and learn from the sign language gestures.