Section: New Software and Platforms

Synthetic human 3D pose dataset

Participants : Gregory Rogez, Cordelia Schmid.

Participants: Gregory Rogez, Cordelia Schmid This large-scale dataset consists of 2,000,000 artificial RGB images of humans and associated 2D and 3D pose annotations. This dataset was generated using the image-based rendering algorithm presented in [19] and has been used to train state-of-the-art Convolutional Neural Networks (CNN) for in-the-wild 3D human pose estimation, see http://www.gregrogez.net/research/human-pose-data-synthesis-for-cnn/.