It contains 90k 3dMD system-captured RGB images and 13k 3dMDhand scans from 263 subjects (131 male, 132 female) with diverse skin tones and hand shapes.
MPI-IS trains SUPR (2022), a full-body, hand, head, and foot model, with 3dMD’s real-world 4D scan data, presented at ECCV 2022.
“Cyber Valley is an example of the attractiveness of Germany as an AI location. Science and industry work hand in hand here and show how the transfer of excellent research into innovative applications and business models can succeed,” said Chancellor Angela Merkel on Wednesday following a virtual tour through research facilities of the Cyber Valley Initiative together with Prime Minister Winfried Kretschmann.
MPI-IS trains its 2020 CAPE generative model for dressing 3D people in clothing with 3dMD’s 4D real-world body data.
Date: January 2020. Source: International Journal of Computer Vision 128, 873–890 (2020). https://doi.org/10.1007/s11263-019-01279-w. Abstract: The optical flow of humans is well known to be useful for the analysis of human action. Recent optical flow methods focus on training deep networks to approach the problem. However, the training data used by them does not cover the…
Date: September 2019. Source: ACM Symposium on Applied Perception 2019, University of Barcelona, Spain. Abstract: The creation of realistic self-avatars that users identify with is important for many virtual reality applications. However, current approaches for creating biometrically plausible avatars that represent a particular individual require expertise and are time-consuming. We investigated the visual perception of…
Date: June 2019. Source: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, USA. Proceedings Page(s): 10093-10103. Abstract: Audio-driven 3D facial animation has been widely explored, but achieving realistic, human-like performance is still unsolved. This is due to the lack of available 3D datasets, models, and standard evaluation metrics. To address…
How SMPL-X combines MANO and FLAME, both trained in part on 3dMD data, into one unified expressive body model from MPI-IS.
MPI-IS builds AMASS (2019), the largest unified mocap archive, with 3dMD’s real-world 4D scan data.
MPI-IS trains CoMA (2018), a mesh autoencoder for generating 3D faces, with 3dMD’s real-world 4D facial data.