BISINDO SIGN LANGUAGE RECOGNITION

Description

This dataset contains images of Indonesian Sign Language (Bahasa Isyarat Indonesia/BISINDO) hand gestures collected for the development and evaluation of computer vision and deep learning-based sign language recognition systems. The dataset consists of annotated hand gesture images representing BISINDO alphabet signs and commonly used gesture classes. Data acquisition was conducted under controlled and semi-controlled environments using RGB cameras with variations in illumination, background conditions, hand orientations, and distances from the camera. The collected images were subsequently preprocessed and organized into class-specific directories to facilitate supervised learning tasks. This dataset is intended to support research in sign language recognition, computer vision, human-computer interaction, assistive technologies, gesture recognition, and deep learning. It can be utilized for image classification, object detection, hand landmark extraction, transfer learning, and multimodal sign language recognition studies. Potential applications include: Automatic BISINDO sign language recognition systems. Human-computer interaction interfaces. Assistive technologies for deaf and hard-of-hearing communities. Deep learning model benchmarking and comparison. Hand gesture detection and classification research. The dataset may be used to train and evaluate machine learning models such as Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), YOLO-based detectors, and hybrid deep learning architectures. Researchers are encouraged to cite this dataset when used in academic publications.

Steps to reproduce

Download and extract the dataset. Review the dataset structure and class labels. Load the images using any machine learning or computer vision framework. Preprocess the images according to the target application. Split the dataset into training, validation, and testing sets. Train a classification or detection model using the provided images. Evaluate the trained model on the testing set using standard performance metrics. Report the experimental configuration and results for reproducibility.

Institutions

• State University of Malang

  East Java, Malang

Categories

Licence