Research map / 3D vision / Depth and stereo
DeepMVS: Learning Multi-View Stereopsis
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use DeepMVS instead of COLMAP (unfiltered) | ETH3D reconstruction · ETH3D | geometric error: 0.046 -> 0.036 | Helped |
| Use DeepMVS instead of COLMAP (unfiltered) | ETH3D reconstruction · ETH3D | photometric error: 0.218 -> 0.224 | Hurt |
| Add U-Net for non-local feature aggregation (ablation) | DeepMVS · ETH3D | geometric error: 0.051 -> 0.043 | Helped |
| Add U-Net for non-local feature aggregation (ablation) | DeepMVS · ETH3D | photometric error: 0.242 -> 0.230 | Helped |
| Add VGG semantic features (ablation) | DeepMVS with U-Net · ETH3D | geometric error: 0.043 -> 0.040 | Helped |
| Add VGG semantic features (ablation) | DeepMVS with U-Net · ETH3D | photometric error: 0.230 -> 0.226 | Helped |
| Add DenseCRF refinement (ablation) | DeepMVS with U-Net and VGG features · ETH3D | geometric error: 0.040 -> 0.036 | Helped |
| Add DenseCRF refinement (ablation) | DeepMVS with U-Net and VGG features · ETH3D | photometric error: 0.226 -> 0.224 | Helped |
| Train with MVS-Synth (ablation) | DeepMVS · ETH3D | geometric error: 0.037 -> 0.036 | Helped |
| Train with MVS-Synth (ablation) | DeepMVS · ETH3D | photometric error: 0.225 -> 0.224 | Helped |
Closest papers by meaning
- MVS2D: Efficient Multi-View Stereo via Attention-Driven 2D Convolutions (CVPR 2022 · similarity 0.92)
- BlendedMVS: A Large-Scale Dataset for Generalized Multi-View Stereo Networks (CVPR 2020 · similarity 0.91)
- Fast-MVSNet: Sparse-to-Dense Multi-View Stereo With Learned Propagation and Gauss-Newton Refinement (CVPR 2020 · similarity 0.90)
- DeepVideoMVS: Multi-View Stereo on Video With Recurrent Spatio-Temporal Fusion (CVPR 2021 · similarity 0.90)
- Learning a Multi-View Stereo Machine (NeurIPS 2017 · similarity 0.90)
- Generalized Binary Search Network for Highly-Efficient Multi-View Stereo (CVPR 2022 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.