S2F2: Self-Supervised High Fidelity Face Reconstruction from Monocular Image

Autor:	Dib, Abdallah, Ahn, Junghyun, Thebault, Cedric, Gosselin, Philippe-Henri, Chevallier, Louis
Rok vydání:	2022
Předmět:	Computer Science - Computer Vision and Pattern Recognition Computer Science - Graphics Computer Science - Machine Learning 68T45 68T07 68U10 68U05 I.4.5 I.4.8 I.4.9 I.3.3 I.2.10
Druh dokumentu:	Working Paper
Popis:	We present a novel face reconstruction method capable of reconstructing detailed face geometry, spatially varying face reflectance from a single monocular image. We build our work upon the recent advances of DNN-based auto-encoders with differentiable ray tracing image formation, trained in self-supervised manner. While providing the advantage of learning-based approaches and real-time reconstruction, the latter methods lacked fidelity. In this work, we achieve, for the first time, high fidelity face reconstruction using self-supervised learning only. Our novel coarse-to-fine deep architecture allows us to solve the challenging problem of decoupling face reflectance from geometry using a single image, at high computational speed. Compared to state-of-the-art methods, our method achieves more visually appealing reconstruction. Comment: 24 Pages, 22 Figures
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2203.07732 Zobrazit plný text záznamu View this record from Arxiv