Real-Time

This dataset contains RGB images of hand signs collected from multiple volunteers under varying lighting conditions and backgrounds using different smartphone cameras. Images were captured in both daylight and dim lighting at multiple angles, using default camera settings with automatic exposure. The dataset is divided into two sections: one for training a detection model (with annotated bounding boxes) and another for testing a recognition model. It is designed to support research in sign language detection and recognition, especially for Bangla hand signs.

本数据集收录了由多名志愿者在多变光照条件与多样背景环境下，通过多款智能手机摄像头采集的手语RGB图像。所有图像均在日光与弱光两种光照场景下以多视角采集，且全程采用默认相机参数与自动曝光模式。本数据集分为两个模块：其一用于训练检测模型（附带标注边界框），其二用于测试识别模型。本数据集旨在为手语检测与识别领域的研究提供支撑，尤其针对孟加拉手语（Bangla hand signs）。