Research map / 3D vision / Depth and stereo
Pyramid Multi-view Stereo Net with Self-adaptive View Aggregation
Hongwei Yi, Zizhuang Wei, Mingyu Ding, Runze Zhang, Yisong Chen, Guoping Wang, Yu-Wing Tai
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add pixel-wise view aggregation (ablation) | MVSNet-style baseline · DTU evaluation | accuracy error (mm): 0.454 -> 0.390 | Improved |
| add voxel-wise view aggregation (ablation) | MVSNet-style baseline · DTU evaluation | accuracy error (mm): 0.454 -> 0.378 | Improved |
| use voxel-wise instead of pixel-wise aggregation (ablation) | VA-MVSNet · DTU evaluation | overall error (mm): 0.379 -> 0.369 | Improved |
| add multi-metric pyramid depth aggregation (MMP) (ablation) | VA-MVSNet with VoxelVA · DTU evaluation | completeness error (mm): 0.359 -> 0.336 | Improved |
| add multi-metric pyramid depth aggregation (MMP) (ablation) | VA-MVSNet with VoxelVA · DTU evaluation | accuracy error (mm): 0.378 -> 0.379 | Hurt |
| add multi-metric pyramid depth aggregation (MMP) (ablation) | VA-MVSNet with PixelVA · DTU evaluation | overall error (mm): 0.379 -> 0.366 | Improved |
| increase test views from 2 to 5 | VA-MVSNet trained on 3 views · DTU evaluation | overall error (mm): 0.441 -> 0.371 | Improved |
| increase test views from 5 to 7 | VA-MVSNet trained on 3 views · DTU evaluation | overall error (mm): 0.371 -> 0.370 | No clear effect |
| train with 5 views instead of 3 | VA-MVSNet (test N=7) · DTU evaluation | overall error (mm): 0.370 -> 0.369 | No clear effect |
| use one pyramid level (K=1) with MMP (ablation) | VA-MVSNet · DTU evaluation | overall error (mm): 0.369 -> 0.361 | Improved |
| increase pyramid levels from K=1 to K=3 (ablation) | PVA-MVSNet · DTU evaluation | completeness error (mm): 0.350 -> 0.336 | Mixed |
| add multi-metric pyramid aggregation and view aggregation vs MVSNet | PVA-MVSNet · Tanks and Temples (no fine-tuning) | mean f-score: 43.48 -> 54.46 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- P-MVSNet — P-MVSNet: Learning Patch-Wise Matching Confidence Aggregation for Multi-View Stereo (ICCV 2019) · better on 3 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Point-Based Multi-View Stereo Network (ICCV 2019 · similarity 0.91)
- Visibility-Aware Point-Based Multi-View Stereo Network (TPAMI 2020 · similarity 0.91)
- Just a Few Points Are All You Need for Multi-View Stereo: A Novel Semi-Supervised Learning Method for Multi-View Stereo (ICCV 2021 · similarity 0.90)
- Dense Hybrid Recurrent Multi-view Stereo Net with Dynamic Consistency Checking (ECCV 2020 · similarity 0.90)
- Mesh-Guided Multi-View Stereo With Pyramid Architecture (CVPR 2020 · similarity 0.90)
- RRT-MVS: Recurrent Regularization Transformer for Multi-View Stereo (AAAI 2025 · similarity 0.88)
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.
