Giuseppe Vecchio, Simone Palazzo and Concetto Spampinato
This is the repo where the official PyTorch implementation for paper "SurfaceNet: Adversarial SVBRDF Estimation from a Single Image" will be released.
Our super trained monkeys 🐒 are working night and day to clean up the code, stay tuned...
In this paper we present SurfaceNet, an approach for estimating spatially-varying bidirectional reflectance distribution function (SVBRDF) material properties from a single image. We pose the problem as an image translation task and propose a novel patch-based generative adversarial network (GAN) that is able to produce high-quality, high-resolution surface reflectance maps. The employment of the GAN paradigm has a twofold objective: 1) allowing the model to recover finer details than standard translation models; 2) reducing the domain shift between synthetic and real data distributions in an unsupervised way.
An extensive evaluation, carried out on a public benchmark of synthetic and real images under different illumination conditions, shows that SurfaceNet largely outperforms existing SVBRDF reconstruction methods, both quantitatively and qualitatively. Furthermore, SurfaceNet exhibits a remarkable ability in generating high-quality maps from real samples without any supervision at training time.
@article{DBLP:journals/corr/abs-2107-11298,
author = {Giuseppe Vecchio and
Simone Palazzo and
Concetto Spampinato},
title = {SurfaceNet: Adversarial {SVBRDF} Estimation from a Single Image},
journal = {CoRR},
volume = {abs/2107.11298},
year = {2021},
url = {https://arxiv.org/abs/2107.11298},
archivePrefix = {arXiv},
eprint = {2107.11298},
timestamp = {Thu, 29 Jul 2021 16:14:15 +0200},
biburl = {https://dblp.org/rec/journals/corr/abs-2107-11298.bib},
bibsource = {dblp computer science bibliography, https://dblp.org}
}