Research map / 3D vision / Depth and stereo
AA-RMVSNet: Adaptive Aggregation Recurrent Multi-View Stereo Network
Zizhuang Wei, Qingtian Zhu, Chen Min, Yisong Chen, Guoping Wang
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 intra-view adaptive aggregation (deformable multi-scale features) to the baseline (ablation) | AA-RMVSNet baseline · DTU evaluation set | overall error (mm): 0.391 -> 0.371 | Improved |
| add intra-view adaptive aggregation to the baseline (ablation) | AA-RMVSNet baseline · DTU evaluation set | completeness error (mm): 0.374 -> 0.346 | Improved |
| add intra-view adaptive aggregation to the baseline (ablation) | AA-RMVSNet baseline · DTU evaluation set | memory (GB): 2.41 -> 4.15 | Hurt |
| add inter-view adaptive aggregation to the baseline (ablation) | AA-RMVSNet baseline · DTU evaluation set | overall error (mm): 0.391 -> 0.370 | Improved |
| add inter-view adaptive aggregation to the baseline (ablation) | AA-RMVSNet baseline · DTU evaluation set | accuracy error (mm): 0.408 -> 0.377 | Improved |
| add inter-view adaptive aggregation to the baseline (ablation) | AA-RMVSNet baseline · DTU evaluation set | memory (GB): 2.41 -> 2.52 | No clear effect |
| combine both intra-view and inter-view aggregation modules (ablation) | AA-RMVSNet baseline · DTU evaluation set | overall error (mm): 0.391 -> 0.357 | Improved |
| increase number of input views N from 3 to 7 (ablation) | AA-RMVSNet · DTU evaluation set | overall error (mm): 0.405 -> 0.386 -> 0.380 | Improved |
| increase depth hypotheses D from 256 to 512 (ablation) | AA-RMVSNet · DTU evaluation set | overall error (mm): 0.380 -> 0.372 | Improved |
| increase input resolution from 480x360 to 800x600 (ablation) | AA-RMVSNet · DTU evaluation set | overall error (mm): 0.372 -> 0.366 -> 0.357 | Improved |
| use recurrent cost regularization with adaptive aggregation instead of 3D CNN (vs MVSNet) | AA-RMVSNet · DTU evaluation set | memory (GB): 15.4 -> 4.25 | Improved |
Later papers that compared with this method
4 later accepted papers report AA-RMVSNet as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Rethinking Disparity: A Depth Range Free Multi-View Stereo Based on Disparity (AAAI 2023 · better on 0 of 1 shared result)
- Efficient Multi-View Stereo by Iterative Dynamic Cost Volume (CVPR 2022 · better on 0 of 1 shared result)
- Non-Parametric Depth Distribution Modelling Based Depth Inference for Multi-View Stereo (CVPR 2022 · better on 0 of 1 shared result)
- MVSTER: Epipolar Transformer for Efficient Multi-View Stereo (ECCV 2022 · better on 1 of 1 shared result)
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
- Recurrent MVSNet for High-Resolution Multi-View Stereo Depth Inference (CVPR 2019 · similarity 0.92)
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching (CVPR 2020 · similarity 0.92)
- RIAV-MVS: Recurrent-Indexing an Asymmetric Volume for Multi-View Stereo (CVPR 2023 · similarity 0.89)
- P-MVSNet: Learning Patch-Wise Matching Confidence Aggregation for Multi-View Stereo (ICCV 2019 · similarity 0.89)
- Pyramid Multi-view Stereo Net with Self-adaptive View Aggregation (ECCV 2020 · similarity 0.89)
- RRT-MVS: Recurrent Regularization Transformer for Multi-View Stereo (AAAI 2025 · similarity 0.87)
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.
