BdSL-126: High-Quality & Annotated Image Dataset; Bangla Sign Language Word Detection Using Real Time Video

BdSL-126: High-Quality & Annotated Image Dataset; Bangla Sign Language Word Detection Using Real-Time Video is a curated and annotated image dataset designed to support research on Bangla Sign Language recognition, word-level sign detection, and real-time video-based sign language interpretation. The dataset consists of image samples representing 126 distinct Bangla Sign Language word classes, making it suitable for the development and evaluation of computer vision and deep learning models.The dataset is systematically organized into 126 primary folders, with each folder corresponding to a specific Bangla Sign Language word. Each primary folder contains two subfolders. The images subfolder includes the original image samples of the corresponding sign word, while the labels subfolder contains the associated annotation files. These annotations are intended to facilitate supervised learning tasks, particularly object detection and recognition using deep learning-based frameworks. This dataset has been prepared to assist researchers, students, and developers in building automated Bangla Sign Language detection systems. Its annotated structure makes it applicable for training, validation, and testing of models such as YOLO and other computer vision architectures. The dataset may also contribute to the development of real-time assistive technologies that improve communication accessibility for Bangla-speaking deaf and hard-of-hearing individuals.The primary objective of BdSL-126 is to provide a high-quality, organized, and annotation-ready resource for advancing Bangla Sign Language recognition research. By offering word-level image data with corresponding labels, this dataset can support further studies in gesture recognition, human-computer interaction, inclusive communication systems, and AI-based assistive technology. Dataset Structure: The dataset contains 126 main folders, each representing one Bangla Sign Language word class. Each folder includes an images directory containing the original sign images and a labels directory containing the corresponding annotation files. Potential Applications: This dataset can be used for Bangla Sign Language word detection, real-time sign language recognition, gesture classification, object detection model training, computer vision research, and the development of assistive communication technologies.