BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Multi-view Neural Surface Reconstruction with Learned Regularizati
 ons
DTSTART:20230630T093000
DTEND:20230630T113000
DTSTAMP:20260407T101423Z
UID:0c11990de7591f29407e2c3f6f88770bcdee54a7c47bd1d6ce50a212
CATEGORIES:Conferences - Seminars
DESCRIPTION:Aoxiang Fan\nEDIC candidacy exam\nExam president: Prof. Sabine
  Süsstrunk\nThesis advisor: Prof. Pascal Fua\nCo-examiner: Prof. Mark Pau
 ly\n\nAbstract\nRecent advances in Multi-view Surface Reconstruction show 
 that a 3D scene representation and learning framework built upon volume re
 ndering and Neural Radiance Fields (NeRF) can significantly improve the re
 construction performance. This fact encourages people to extend the approa
 ch to include more sophisticated physically-based formulations. On the oth
 er aspect\, following the conventional Structure-from-Motion (SfM) pipelin
 e\, it has also been shown that 3D geometry can be learned with a data-dri
 ven approach from 2D image observations regardless of the physically-based
  formulations.\n\nFollowing a comprehensive and meticulous examination of 
 the literature\, our proposed research focuses on three aspects: 1) extend
 ing physically-based models in the context of inverse rendering to facilit
 ate Multi-view Surface Reconstruction 2) studying the role of data-driven 
 priors in physically-based formulations to make the ill-posed problem effi
 ciently solvable\; 3) investigating effective architectures and data-effic
 ient training paradigms to learn better image representation for data-driv
 en Multi-view Surface Reconstruction.\n\nBackground papers\na. NeuS: Lear
 ning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstru
 ction. Peng Wang et al\, NeurIPS 2021\, https://proceedings.neurips.cc/pa
 per/2021/file/e41e164f7485ec4a28741a2d0ea41c74-Paper.pdf\n\nb. TransMVSNe
 t: Global Context-aware Multi-view Stereo Network with Transformers. Yik
 ang Ding et al\, CVPR 2022\, http://openaccess.thecvf.com/content/CVPR202
 2/papers/Ding_TransMVSNet_Global_Context-Aware_Multi-View_Stereo_Network_W
 ith_Transformers_CVPR_2022_paper.pdf\n\nc. VolRecon: Volume Rendering of 
 Signed Ray Distance Functions for Generalizable Multi-View Reconstruction
 . Yufan Ren et al\, CVPR 2023\, https://arxiv.org/pdf/2212.08067 \n 
LOCATION:BC 229 https://plan.epfl.ch/?room==BC%20229
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR