We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CV

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computer Vision and Pattern Recognition

Title: A 3D Mesh-based Lifting-and-Projection Network for Human Pose Transfer

Abstract: Human pose transfer has typically been modeled as a 2D image-to-image translation problem. This formulation ignores the human body shape prior in 3D space and inevitably causes implausible artifacts, especially when facing occlusion. To address this issue, we propose a lifting-and-projection framework to perform pose transfer in the 3D mesh space. The core of our framework is a foreground generation module, that consists of two novel networks: a lifting-and-projection network (LPNet) and an appearance detail compensating network (ADCNet). To leverage the human body shape prior, LPNet exploits the topological information of the body mesh to learn an expressive visual representation for the target person in the 3D mesh space. To preserve texture details, ADCNet is further introduced to enhance the feature produced by LPNet with the source foreground image. Such design of the foreground generation module enables the model to better handle difficult cases such as those with occlusions. Experiments on the iPER and Fashion datasets empirically demonstrate that the proposed lifting-and-projection framework is effective and outperforms the existing image-to-image-based and mesh-based methods on human pose transfer task in both self-transfer and cross-transfer settings.
Subjects: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
Cite as: arXiv:2109.11719 [cs.CV]
  (or arXiv:2109.11719v1 [cs.CV] for this version)

Submission history

From: Yangheng Zhao [view email]
[v1] Fri, 24 Sep 2021 03:03:02 GMT (5379kb,D)

Link back to: arXiv, form interface, contact.