MVSFormer: Multi-View Stereo by Learning Robust Image Features and Temperature-based Depth

Published: 20 Dec 2022, Last Modified: 28 Feb 2023Accepted by TMLREveryoneRevisionsBibTeX
Abstract: Feature representation learning is the key recipe for learning-based Multi-View Stereo (MVS). As the common feature extractor of learning-based MVS, vanilla Feature Pyramid Networks (FPNs) suffer from discouraged feature representations for reflection and texture-less areas, which limits the generalization of MVS. Even FPNs worked with pre-trained Convolutional Neural Networks (CNNs) fail to tackle these issues. On the other hand, Vision Transformers (ViTs) have achieved prominent success in many 2D vision tasks. Thus we ask whether ViTs can facilitate feature learning in MVS? In this paper, we propose a pre-trained ViT enhanced MVS network called MVSFormer, which can learn more reliable feature representations benefited by informative priors from ViT. The finetuned MVSFormer with hierarchical ViTs of efficient attention mechanisms can achieve prominent improvement based on FPNs. Besides, the alternative MVSFormer with frozen ViT weights is further proposed. This largely alleviates the training cost with competitive performance strengthened by the attention map from the self-distillation pre-training. MVSFormer can be generalized to various input resolutions with efficient multi-scale training strengthened by gradient accumulation. Moreover, we discuss the merits and drawbacks of classification and regression-based MVS methods, and further propose to unify them with a temperature-based strategy. MVSFormer achieves state-of-the-art performance on the DTU dataset. Particularly, MVSFormer ranks as Top-1 on both intermediate and advanced sets of the highly competitive Tanks-and-Temples leaderboard. Codes and models are released in
Submission Length: Regular submission (no more than 12 pages of main content)
Changes Since Last Submission: Compared with the last TMLR submission, we squeeze the ETH3D experiments of previous Appendix.C into the main paper, and revise some details about the experiment introduction. We also provide author information, a project GitHub link, and acknowledgment.
Assigned Action Editor: ~David_Fouhey2
License: Creative Commons Attribution 4.0 International (CC BY 4.0)
Submission Number: 443