DHP19 is a dataset of 3D human poses collected from DVS event cameras. The dataset uses event cameras at 4 different locations, and the output of each event camera at different locations is recorded. A motion capture system is used to obtain the true pose of human motion. The 3D coordinates are extracted from Vicon, while the 2D coordinates are manually annotated. The dataset records the event streams and 3D positions of 17 subjects, each of whom performed 33 actions.