Self-Supervised 3D Representation Learning of Dressed Humans From Social Media Videos

Yasamin Jafarian, Hyun Soo Park

Research output: Contribution to journalArticlepeer-review

Abstract

A key challenge of learning a visual representation for the 3D high fidelity geometry of dressed humans lies in the limited availability of the ground truth data (e.g., 3D scanned models), which results in the performance degradation of 3D human reconstruction when applying to real-world imagery. We address this challenge by leveraging a new data resource: a number of social media dance videos that span diverse appearance, clothing styles, performances, and identities. Each video depicts dynamic movements of the body and clothes of a single person while lacking the 3D ground truth geometry. To learn a visual representation from these videos, we present a new self-supervised learning method to use the local transformation that warps the predicted local geometry of the person from an image to that of another image at a different time instant. This allows self-supervision by enforcing a temporal coherence over the predictions. In addition, we jointly learn the depths along with the surface normals that are highly responsive to local texture, wrinkle, and shade by maximizing their geometric consistency. Our method is end-to-end trainable, resulting in high fidelity depth estimation that predicts fine geometry faithful to the input real image. We further provide a theoretical bound of self-supervised learning via an uncertainty analysis that characterizes the performance of the self-supervised learning without training. We demonstrate that our method outperforms the state-of-the-art human depth estimation and human shape recovery approaches on both real and rendered images.

Original languageEnglish (US)
Pages (from-to)8969-8983
Number of pages15
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume45
Issue number7
DOIs
StatePublished - Jul 1 2023

Bibliographical note

Publisher Copyright:
IEEE

Keywords

  • Depth estimation
  • dataset
  • high fidelity human reconstruction
  • normal estimation
  • self-supervised learning
  • single view 3D reconstruction

PubMed: MeSH publication types

  • Journal Article
  • Research Support, U.S. Gov't, Non-P.H.S.

Fingerprint

Dive into the research topics of 'Self-Supervised 3D Representation Learning of Dressed Humans From Social Media Videos'. Together they form a unique fingerprint.

Cite this