Research map / 3D vision / Depth and stereo
Multi-View Stereo Representation Revist: Region-Aware MVSNet
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 |
|---|---|---|---|
| add SDF prediction branch to the depth-only model (ablation) | CasMVSNet · DTU evaluation set | Overall (mm): 0.319 -> 0.310 | Helped |
| fuse SDF predictions with the depth branch (ablation) | two-branch MVS model · DTU evaluation set | Overall (mm): 0.310 -> 0.302 | Helped |
| add SDF prediction branch to the depth-only model (ablation) | CasMVSNet · Tanks and Temples | F-score: 64.02 -> 64.62 | Helped |
| fuse SDF predictions with the depth branch (ablation) | two-branch MVS model · Tanks and Temples | F-score: 64.62 -> 65.39 | Helped |
| use RFP instead of FPN as the image encoder (ablation) | RA-MVSNet · DTU evaluation set | Overall (mm): 0.302 -> 0.297 | Helped |
| increase fusion threshold θ (ablation) | RA-MVSNet · DTU evaluation set | Overall (mm): 0.297 -> 0.302 -> 0.310 -> 0.310 | Hurt |
| increase inference image resolution (ablation) | RA-MVSNet · Tanks and Temples | F-score: 57.69 -> 63.62 -> 65.72 | Helped |
| increase inference image resolution (ablation) | RA-MVSNet · Tanks and Temples | Mem.: 3702M -> 6799M -> 10527M | Hurt |
| increase local patch size up to 5 (ablation) | RA-MVSNet signed-distance ground-truth search · BlendedMVS and DTU | reconstruction performance | Helped |
| use patch size larger than 5 (ablation) | RA-MVSNet signed-distance ground-truth search · BlendedMVS and DTU | reconstruction performance | No clear effect |
| increase the number of input images (ablation) | RA-MVSNet · DTU | reconstruction performance | Mixed |
| add the SDF branch (ablation) | CasMVSNet · DTU evaluation set | Memory: 8487M -> 8624M | Hurt |
Closest papers by meaning
- GeoMVSNet: Learning Multi-View Stereo With Geometry Perception (CVPR 2023 · similarity 0.90)
- Mesh-Guided Multi-View Stereo With Pyramid Architecture (CVPR 2020 · similarity 0.90)
- Dual-Level Precision Edges Guided Multi-View Stereo with Accurate Planarization (AAAI 2025 · similarity 0.89)
- MVSFormer++: Revealing the Devil in Transformer's Details for Multi-View Stereo (ICLR 2024 · similarity 0.88)
- RRT-MVS: Recurrent Regularization Transformer for Multi-View Stereo (AAAI 2025 · similarity 0.88)
- DVP-MVS: Synergize Depth-Edge and Visibility Prior 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.