mlsanityData to 7 Oct 2026

MVS2D: Efficient Multi-view Stereo via Attention-Driven 2D Convolutions

arXiv 2104.13325 ↗First posted 27 Apr 2021cs.CVcited by 43

Not on the map. Why?

Abstract (arXiv)

Deep learning has made significant impacts on multi-view stereo systems. State-of-the-art approaches typically involve building a cost volume, followed by multiple 3D convolution operations to recover the input image's pixel-wise depth. While such end-to-end learning of plane-sweeping stereo advances public benchmarks' accuracy, they are typically very slow to compute. We present ouralg, a highly efficient multi-view stereo algorithm that seamlessly integrates multi-view constraints into single-view networks via an attention mechanism. Since ouralg only builds on 2D convolutions, it is at least 2× faster than all the notable counterparts. Moreover, our algorithm produces precise depth estimations and 3D reconstructions, achieving state-of-the-art results on challenging benchmarks ScanNet, SUN3D, RGBD, and the classical DTU dataset. our algorithm also out-performs all other algorithms in the setting of inexact camera poses. Our code is released at urlhttps://github.com/zhenpeiyang/MVS2D

Key facts. Cited by 43 papers in the map (data through 7 Oct 2026). Builds on 39 of its 61 references that are in the map. First posted on arXiv 27 Apr 2021.

Builds on

39 of 61 references.

References · 39

  1. 2017Attention Is All You Needcited by 59,932
  2. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  3. 2017ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenescited by 3,088
  4. 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,335
  5. 2017Non-local Neural Networkscited by 2,290
  6. 2018Self-Attention with Relative Position Representationscited by 1,112
  7. 2017PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volumecited by 966
  8. 2015A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimationcited by 878
  9. 2018MVSNet: Depth Inference for Unstructured Multi-view Stereocited by 780
  10. 2019Stand-Alone Self-Attention in Vision Modelscited by 542
  11. 2019BlendedMVS: A Large-scale Dataset for Generalized Multi-view Stereo Networkscited by 487
  12. 2019Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matchingcited by 414
27 more
  1. 2019From Big to Small: Multi-Scale Local Planar Guidance for Monocular Depth Estimationcited by 379
  2. 2018DeepMVS: Learning Multi-view Stereopsiscited by 277
  3. 2019Recurrent MVSNet for High-resolution Multi-view Stereo Depth Inferencecited by 252
  4. 2019Enforcing geometric constraints of virtual normal for depth predictioncited by 233
  5. 2020Atlas: End-to-End 3D Scene Reconstruction from Posed Imagescited by 206
  6. 2020PatchmatchNet: Learned Multi-View Patchmatch Stereocited by 204
  7. 2020AANet: Adaptive Aggregation Network for Efficient Stereo Matchingcited by 202
  8. 2016DeMoN: Depth and Motion Network for Learning Monocular Stereocited by 200
  9. 2017Learning a Multi-View Stereo Machinecited by 194
  10. 2019Cost Volume Pyramid Based Depth Inference for Multi-View Stereocited by 154
  11. 2019Deep Stereo using Adaptive Thin Volume Representation with Uncertainty Awarenesscited by 151
  12. 2019Point-Based Multi-View Stereo Networkcited by 147
  13. 2017SurfaceNet: An End-to-end 3D Neural Network for Multiview Stereopsiscited by 128
  14. 2020HITNet: Hierarchical Iterative Tile Refinement Network for Real-time Stereo Matchingcited by 120
  15. 2019DPSNet: End-to-end Deep Plane Sweep Stereocited by 115
  16. 2019DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatchcited by 114
  17. 2020Fast-MVSNet: Sparse-to-Dense Multi-View Stereo With Learned Propagation and Gauss-Newton Refinementcited by 105
  18. 2014Computing the Stereo Matching Cost with a Convolutional Neural Networkcited by 74
  19. 2017Multi-Scale Continuous CRFs as Sequential Deep Networks for Monocular Depth Estimationcited by 70
  20. 2019Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volumecited by 59
  21. 2020Multi-view Depth Estimation using Epipolar Spatio-Temporal Networkscited by 59
  22. 2019Normal Assisted Stereo Depth Estimationcited by 49
  23. 2020Occlusion-Aware Depth Estimation with Adaptive Normal Constraintscited by 41
  24. 2018Learning Spatial Common Sense with Geometry-Aware Recurrent Networkscited by 35
  25. 2020Bi3D: Stereo Depth Estimation via Binary Classificationscited by 34
  26. 2018Geometry-Aware Recurrent Neural Networks for Active Visual Recognitioncited by 20
  27. 2019Geometry-Aware Neural Renderingcited by 10

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 4; 2022: 6; 2023: 9; 2024: 9; 2025: 12; 2026: 3.

43 papers.

Sort:

By area:

  1. 2023DUSt3R: Geometric 3D Vision Made Easycited by 1,526
  2. 2024DepthSplat: Connecting Gaussian Splatting and Depthcited by 275
  3. 2023Iterative Geometry Encoding Volume for Stereo Matchingcited by 188
  4. 2022Is Attention All That NeRF Needs?cited by 113
  5. 2025Pow3R: Empowering Unconstrained 3D Reconstruction with Camera and Scene Priorscited by 69
  6. 2024World-consistent Video Diffusion with Explicit 3D Modelingcited by 62
  7. 2022MVSTER: Epipolar Transformer for Efficient Multi-View Stereocited by 50
  8. 2022MVSFormer: Multi-View Stereo by Learning Robust Image Features and Temperature-based Depthcited by 50
  9. 2025Matrix3D: Large Photogrammetry Model All-in-Onecited by 39
  10. 2021VoRTX: Volumetric 3D Reconstruction With Transformers for Voxelwise View Selection and Fusioncited by 38
  11. 2022A Benchmark and a Baseline for Robust Multi-view Depth Estimationcited by 38
  12. 2021Generalized Binary Search Network for Highly-Efficient Multi-View Stereocited by 32
31 more
  1. 2021Input-level Inductive Biases for 3D Reconstructioncited by 31
  2. 2024Adaptive Fusion of Single-View and Multi-View Depth for Autonomous Drivingcited by 30
  3. 2024Cross-View Completion Models are Zero-shot Correspondence Estimatorscited by 30
  4. 2025MVSAnywhere: Zero-Shot Multi-View Stereocited by 26
  5. 2023Learning to Fuse Monocular and Multi-view Cues for Multi-frame Depth Estimation in Dynamic Scenescited by 25
  6. 20213DVNet: Multi-View Depth Prediction and Volumetric Refinementcited by 21
  7. 2022WT-MVSNet: Window-based Transformers for Multi-view Stereocited by 20
  8. 2025MonSter++: Unified Stereo Matching, Multi-view Stereo, and Real-time Stereo with Monodepth Priorscited by 19
  9. 2025Test3R: Learning to Reconstruct 3D at Test Timecited by 19
  10. 2025Multi-view Reconstruction via SfM-guided Monocular Depth Estimationcited by 17
  11. 2023RayMVSNet++: Learning Ray-based 1D Implicit Fields for Accurate Multi-View Stereocited by 15
  12. 2023Free-Editor: Zero-shot Text-driven 3D Scene Editingcited by 14
  13. 2025PE3R: Perception-Efficient 3D Reconstructioncited by 9
  14. 2023Joint Depth Prediction and Semantic Segmentation with Multi-View SAMcited by 8
  15. 2024Direct and Explicit 3D Generation from a Single Imagecited by 6
  16. 2024Pixel-Aligned Multi-View Generation with Depth Guided Decodercited by 4
  17. 2025Multi-view Gaze Target Estimationcited by 4
  18. 2022Rethinking Disparity: A Depth Range Free Multi-View Stereo Based on Disparitycited by 3
  19. 2023Uncertainty-Driven Dense Two-View Structure from Motioncited by 3
  20. 2024SceneFactory: A Workflow-centric and Unified Framework for Incremental Scene Modelingcited by 3
  21. 2025MoRe: Monocular Geometry Refinement via Graph Optimization for Cross-View Consistencycited by 3
  22. 2023Multi-View Stereo Representation Revisit: Region-Aware MVSNetcited by 2
  23. 2023CT-MVSNet: Efficient Multi-View Stereo with Cross-scale Transformercited by 2
  24. 2024MonoPatchNeRF: Improving Neural Radiance Fields with Patch-based Monocular Guidancecited by 2
  25. 2025Towards Depth Foundation Model: Recent Trends in Vision-Based Depth Estimationcited by 2
  26. 2024SCENES: Subpixel Correspondence Estimation With Epipolar Supervisioncited by 1
  27. 2025MuGS: Multi-Baseline Generalizable Gaussian Splatting Reconstructioncited by 1
  28. 2026Enhancing MLLM Spatial Understanding via Active 3D Scene Exploration for Multi-Perspective Reasoningcited by 1
  29. 2025Detail-aware multi-view stereo network for depth estimationcited by 0
  30. 2026PAGaS: Pixel-Aligned 1DoF Gaussian Splatting for Depth Refinementcited by 0
  31. 2026ZipMVS: Multi-View Stereo with Compressed Cost Volumescited by 0