English

RGB-D-Fusion: Image Conditioned Depth Diffusion of Humanoid Subjects

Computer Vision and Pattern Recognition 2023-09-25 v1 Machine Learning

Abstract

We present RGB-D-Fusion, a multi-modal conditional denoising diffusion probabilistic model to generate high resolution depth maps from low-resolution monocular RGB images of humanoid subjects. RGB-D-Fusion first generates a low-resolution depth map using an image conditioned denoising diffusion probabilistic model and then upsamples the depth map using a second denoising diffusion probabilistic model conditioned on a low-resolution RGB-D image. We further introduce a novel augmentation technique, depth noise augmentation, to increase the robustness of our super-resolution model.

Keywords

Cite

@article{arxiv.2307.15988,
  title  = {RGB-D-Fusion: Image Conditioned Depth Diffusion of Humanoid Subjects},
  author = {Sascha Kirch and Valeria Olyunina and Jan Ondřej and Rafael Pagés and Sergio Martin and Clara Pérez-Molina},
  journal= {arXiv preprint arXiv:2307.15988},
  year   = {2023}
}