UBC3V Dataset

~6 million synthetic depth frames for pose estimation from multiple cameras.

Source: Real-Time Human Motion Capture with Multiple Depth Cameras