Research map / 3D vision / Depth and stereo
Learning Neural Implicit through Volume Rendering with Attentive Depth Fusion Priors
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 attentive depth-fusion priors to neural implicit reconstruction | Neural implicit reconstruction · ScanNet | F-score: 0.733 -> 0.902 | Helped |
| add attentive depth-fusion priors to neural implicit reconstruction | Neural implicit reconstruction · Replica | Comp.: 3.00 -> 2.45 | Helped |
| use estimated rather than GT camera poses (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 2.73 -> 3.01 | Hurt |
| use recent-frame online TSDF fusion instead of offline fusion of all depth images (ablation) | Attentive depth-fusion reconstruction with GT poses · Replica | Comp.: 2.45 -> 2.36 | Helped |
| remove the attention mechanism (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 1.44 -> 1.86 | Hurt |
| apply depth-prior attention to low-frequency surfaces instead of high-frequency geometry (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 1.44 -> 2.75 | Hurt |
| apply depth-prior attention to both low- and high-frequency geometry (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 1.44 -> 2.96 | Hurt |
| remove the TSDF bandwidth restriction (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 1.44 -> 1.87 | Hurt |
| increase TSDF bandwidth from 5 to 7 (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 1.44 -> 2.01 | Hurt |
| use high-frequency rather than low-frequency geometry outside the TSDF bandwidth (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 1.44 -> 124.21 | Hurt |
| train rather than fix the low-frequency prediction-prior decoder (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 1.44 -> 43.46 | Hurt |
| remove fixed prediction-prior decoder parameters (ablation) | Attentive depth-fusion reconstruction · Replica | DepthL1: 1.44 -> 125.95 | Hurt |
Closest papers by meaning
- Towards Unbiased Volume Rendering of Neural Implicit Surfaces With Geometry Priors (CVPR 2023 · similarity 0.91)
- NeAT: Learning Neural Implicit Surfaces With Arbitrary Topologies From Multi-View Images (CVPR 2023 · similarity 0.89)
- Volume Rendering of Neural Implicit Surfaces (NeurIPS 2021 · similarity 0.89)
- BNV-Fusion: Dense 3D Reconstruction Using Bi-Level Neural Volume Fusion (CVPR 2022 · similarity 0.89)
- MVD-Fusion: Single-view 3D via Depth-consistent Multi-view Generation (CVPR 2024 · similarity 0.86)
- Sensing Surface Patches in Volume Rendering for Inferring Signed Distance Functions (AAAI 2025 · similarity 0.86)
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.