BDSL49: A Comprehensive Dataset of Bengali Sign Language

This dataset contains images of Bengali sign language. This collection of images represents the Bengali sign language. 14 Bangladeshi adults created and annotated the dataset. The existing Bengali sign language datasets lack adequate and diverse image samples to develop real-life expert systems. As a result, a dataset encompassing all facets of real-world usable signs is necessary. This results in the creation of a comprehensive dataset for Bengali sign language. The proposed dataset contains two sections. The first section involves training a detection model, while the second involves training a recognition model. Both sections feature a wide variety of images taken in a variety of backgrounds and lighting circumstances, with the first section featuring a full-frame image with annotation and the second featuring a cropped image with only the signs. Each section contains 14,714 images separated into 49 classes, for a total of 29,428 images in the dataset.本数据集包含孟加拉语手语（Bengali sign language）图像。该图像集合完整呈现了孟加拉语手语体系。本次数据集由14名孟加拉国成年人完成制作与标注工作。现有孟加拉语手语数据集缺乏足够且多样的图像样本，无法支撑现实场景下专家系统的研发，因此亟需一套覆盖真实可用手语全维度的数据集，由此催生了本套综合性孟加拉语手语数据集。本数据集包含两个部分：第一部分用于训练检测模型，第二部分用于训练识别模型。两部分均涵盖了多种背景与光照条件下拍摄的多样化图像；其中第一部分为带标注的全画幅图像，第二部分为仅保留手语区域的裁剪图像。每个部分均包含14714张图像，分为49个类别，全数据集总图像量达29428张。