This dataset contains three persons performing static signs of American Sign Language alphabet. Each top-level folder contains the recording made for one subject; under top-level folder, there are folders containing recordings for each letter. When using the dataset, please cite: Alina Kuznetsova, Laura Leal-Taixé, Bodo Rosenhahn "Real-time sign language recognition using a consumer depth camera" IEEE International Conference on Computer Vision Workshops (ICCVW), 3rd Workshop on Consumer Depth Cameras for Computer Vision (CDC4CV), December 2013 For each letter, there the following files available: imDepthOrig*_*.png --- 16bit original depth values, delievered by the sensor; imDepthConf*_*.png --- 16bit confidence values [see documentation of Intel Creative Depth Sensor] imDepthMask*_*.png --- 8bit binary hand segmentation [created automatically using Hand detection and segmentation available at http://www.tnt.uni-hannover.de/project/handposerecognition/] imRGB*_*.png --- 24bit RGB images delivered by the depth sensor. The numeration is the following: [take number]_[frame number] {take number} starts with 1; {frame number} starts with 1 and ends by 4 or 5; During the recordings, several frames were recorded sequentially, so the images with the same {take number} and different {frame number} will be quite similar. Please note that Depth and RGB images are not aligned. Use Intel Calibration tool to find the mapping between RGB and depth images. If you have any questions, please contact kuznetso@tnt.uni-hannover.de