Published on Tue Apr 21 2020

M^3VSNet: Unsupervised Multi-metric Multi-view Stereo Network

Baichuan Huang, Hongwei Yi, Can Huang, Yijia He, Jingbin Liu, Xiao Liu

The present Multi-view stereo (MVS) methods with supervised learning-based networks have an impressive performance comparing with traditional MVS methods. We propose a novel multi-metric function that combines pixel-wise and feature-wise loss function to learn the inherent constraints from different perspectives.

0
0
0
Abstract

The present Multi-view stereo (MVS) methods with supervised learning-based networks have an impressive performance comparing with traditional MVS methods. However, the ground-truth depth maps for training are hard to be obtained and are within limited kinds of scenarios. In this paper, we propose a novel unsupervised multi-metric MVS network, named M^3VSNet, for dense point cloud reconstruction without any supervision. To improve the robustness and completeness of point cloud reconstruction, we propose a novel multi-metric loss function that combines pixel-wise and feature-wise loss function to learn the inherent constraints from different perspectives of matching correspondences. Besides, we also incorporate the normal-depth consistency in the 3D point cloud format to improve the accuracy and continuity of the estimated depth maps. Experimental results show that M3VSNet establishes the state-of-the-arts unsupervised method and achieves comparable performance with previous supervised MVSNet on the DTU dataset and demonstrates the powerful generalization ability on the Tanks and Temples benchmark with effective improvement. Our code is available at https://github.com/whubaichuan/M3VSNet.

Thu Apr 30 2020
Computer Vision
M^3VSNet: Unsupervised Multi-metric Multi-view Stereo Network
The present Multi-view stereo (MVS) methods with supervised learning-based networks have an impressive performance comparing with traditional MVS methods. We propose a novel multi-metric function that combines pixel-wise and feature-wise loss function to learn the inherent constraints from different perspectives.
0
0
0
Mon Aug 12 2019
Computer Vision
Point-Based Multi-View Stereo Network
Point-MVSNet is a novel point-based deep framework for multi-view stereo (MVS) It processes the target scene as point clouds and predicts the depth in a coarse-to-fine manner.
0
0
0
Fri Dec 06 2019
Computer Vision
Pyramid Multi-view Stereo Net with Self-adaptive View Aggregation
We propose an effective and efficient pyramid multi-view stereo stereo(MVS) net with self-adaptive view aggregation for accurate and complete dense point cloud reconstruction. Different from using mean square variance to generate cost volume in previous deep-learning based MVS methods.
0
0
0
Fri Sep 25 2020
Machine Learning
Towards General Purpose Geometry-Preserving Single-View Depth Estimation
Single-view depth estimation (SVDE) plays a crucial role in scene Understanding for AR applications, 3D modeling, and robotics. Recent works have shown that a successful solution strongly relies on the diversity and volume of training data.
0
0
0
Tue May 26 2020
Machine Learning
SurfaceNet+: An End-to-end 3D Neural Network for Very Sparse Multi-view Stereopsis
SurfaceNet+ is a volumetric method to handle the 'incompleteness' and 'inaccuracy' problems induced by a very sparse MVS setup. SurfaceNet+ has superiority in selecting valid views while discarding invalid occluded views.
0
0
0
Tue Jul 21 2020
Computer Vision
Dense Hybrid Recurrent Multi-view Stereo Net with Dynamic Consistency Checking
0
0
0