Saved in:
Bibliographic Details
Main Authors: Zhu, Jie, Peng, Bo, Zhang, Zhe, Liu, Bingzheng, Lei, Jianjun
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.11845
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917986746499072
author Zhu, Jie
Peng, Bo
Zhang, Zhe
Liu, Bingzheng
Lei, Jianjun
author_facet Zhu, Jie
Peng, Bo
Zhang, Zhe
Liu, Bingzheng
Lei, Jianjun
contents Learning-based Multi-View Stereo (MVS) methods have made remarkable progress in recent years. However, how to effectively train the network without using real-world labels remains a challenging problem. In this paper, driven by the recent advancements of vision foundation models, a novel method termed DFM-MVS, is proposed to leverage the depth foundation model to generate the effective depth prior, so as to boost MVS in the absence of real-world labels. Specifically, a depth prior-based pseudo-supervised training mechanism is developed to simulate realistic stereo correspondences using the generated depth prior, thereby constructing effective supervision for the MVS network. Besides, a depth prior-guided error correction strategy is presented to leverage the depth prior as guidance to mitigate the error propagation problem inherent in the widely-used coarse-to-fine network structure. Experimental results on DTU and Tanks & Temples datasets demonstrate that the proposed DFM-MVS significantly outperforms existing MVS methods without using real-world labels.
format Preprint
id arxiv_https___arxiv_org_abs_2504_11845
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Boosting Multi-View Stereo with Depth Foundation Model in the Absence of Real-World Labels
Zhu, Jie
Peng, Bo
Zhang, Zhe
Liu, Bingzheng
Lei, Jianjun
Computer Vision and Pattern Recognition
Learning-based Multi-View Stereo (MVS) methods have made remarkable progress in recent years. However, how to effectively train the network without using real-world labels remains a challenging problem. In this paper, driven by the recent advancements of vision foundation models, a novel method termed DFM-MVS, is proposed to leverage the depth foundation model to generate the effective depth prior, so as to boost MVS in the absence of real-world labels. Specifically, a depth prior-based pseudo-supervised training mechanism is developed to simulate realistic stereo correspondences using the generated depth prior, thereby constructing effective supervision for the MVS network. Besides, a depth prior-guided error correction strategy is presented to leverage the depth prior as guidance to mitigate the error propagation problem inherent in the widely-used coarse-to-fine network structure. Experimental results on DTU and Tanks & Temples datasets demonstrate that the proposed DFM-MVS significantly outperforms existing MVS methods without using real-world labels.
title Boosting Multi-View Stereo with Depth Foundation Model in the Absence of Real-World Labels
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.11845