UT-Kinect stands for UTKinect-Action3D Dataset, which is a dataset used to recognize actions from depth sequences.

Data is collected at a frame rate of 15 fps using a fixed Kinect and the depth camera of the Kinect for Windows SDK Beta version, including RGB, Depth, and 3D skeleton data.

The UT-Kinect dataset divides samples into 10 daily life behaviors (i.e., walk, sit down, stand up, pick up, carry, throw, push, pull, wave hands, clap hands), involving 10 subjects, each of whom performs each action twice.