Research map / 3D vision / Depth and stereo
VolumeFusion: Deep Depth Fusion for 3D Scene Reconstruction
Jaesung Choe, Sunghoon Im, Francois Rameau, Minjun Kang, In So Kweon
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 PosedConv instead of standard 3D Conv (ablation) | VolumeFusion · ScanNet | AbsRel: .058 -> .049 | Improved |
| add PosedConv instead of standard 3D Conv (ablation) | VolumeFusion · ScanNet | F-score: .460 -> .508 | Improved |
| use discrete kernel rotation instead of rotation-by-interpolation (ablation) | PosedConv · ScanNet | RMSE: .223 -> .164 | Improved |
| rotate kernels by naive interpolation instead of using standard 3D Conv (ablation) | VolumeFusion · ScanNet | F-score: .460 -> .487 | Improved |
| use two-stage depth fusion instead of single-stage TSDF regression (ablation) | Murez et al. single-stage · ScanNet | RMSE: .248 -> .164 | Improved |
| embed masked depth maps in addition to PosedConv features (ablation) | VolumeFusion w/o depth · ScanNet | TSDF L1: .159 -> .141 | Improved |
| add PosedConv without depth fusion over single-stage (ablation) | Murez et al. · ScanNet | RMSE: .248 -> .236 | Improved |
| add overlapping mask (ablation) | VolumeFusion · ScanNet | AbsRel: .060 -> .049 | Improved |
| add overlapping mask (ablation) | VolumeFusion · ScanNet | F-score: .475 -> .508 | Improved |
| add PosedConv alone without depth fusion or mask (ablation) | Single-stage baseline · ScanNet | AbsRel: .061 -> .060 | No clear effect |
| use VolumeFusion instead of Murez et al. | single-stage TSDF regression · ScanNet | AbsRel: .061 -> .049 | Improved |
| use VolumeFusion instead of Murez et al. | single-stage TSDF regression · ScanNet | F-score: .499 -> .508 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DPSNet — DPSNet: End-to-end Deep Plane Sweep Stereo (ICLR 2019) · better on 3 of 3 shared results
- NeuralRecon — NeuralRecon: Real-Time Coherent 3D Reconstruction From Monocular Video (CVPR 2021) · better on 2 of 2 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
- StereoDRNet: Dilated Residual StereoNet (CVPR 2019 · similarity 0.90)
- Learning Non-Volumetric Depth Fusion Using Successive Reprojections (CVPR 2019 · similarity 0.90)
- NeuralFusion: Online Depth Fusion in Latent Space (CVPR 2021 · similarity 0.90)
- Learning Neural Implicit through Volume Rendering with Attentive Depth Fusion Priors (NeurIPS 2023 · similarity 0.89)
- Atlas: End-to-End 3D Scene Reconstruction from Posed Images (ECCV 2020 · similarity 0.89)
- Multi-view Reconstruction via SfM-guided Monocular Depth Estimation (CVPR 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.
