Research map / 3D vision / Face and shape
Multi-View Supervision for Single-View Reconstruction via Differentiable Ray Consistency
Check 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 |
|---|---|---|---|
| Use DRC instead of depth fusion with noisy depth supervision | ShapeNet reconstruction · ShapeNet airplane | mean IoU: 0.46 -> 0.51 | Helped |
| Use DRC instead of depth fusion with noisy depth supervision | ShapeNet reconstruction · ShapeNet car | mean IoU: 0.71 -> 0.74 | Helped |
| Use DRC instead of depth fusion with noisy depth supervision | ShapeNet reconstruction · ShapeNet chair | mean IoU: 0.39 -> 0.45 | Helped |
| Use depth rather than mask observations (ablation) | DRC · ShapeNet, 1-2 training views | mean IoU | Helped |
| Increase the number of training views per instance (ablation) | DRC · ShapeNet | mean IoU | Mixed |
| Use depth rather than mask observations (ablation) | DRC · ShapeNet, more than 5 training views | mean IoU | No clear effect |
| Add PASCAL view-consistency fine-tuning after ShapeNet 3D pretraining (ablation) | Single-view reconstruction · PASCAL VOC mean | mean IoU: 0.51 -> 0.54 | Helped |
| Add PASCAL view-consistency fine-tuning after ShapeNet 3D pretraining (ablation) | Single-view reconstruction · PASCAL VOC airplane | mean IoU: 0.53 -> 0.55 | Helped |
| Add PASCAL view-consistency fine-tuning after ShapeNet 3D pretraining (ablation) | Single-view reconstruction · PASCAL VOC car | mean IoU: 0.67 -> 0.72 | Helped |
| Add PASCAL view-consistency fine-tuning after ShapeNet 3D pretraining (ablation) | Single-view reconstruction · PASCAL VOC chair | mean IoU: 0.33 -> 0.34 | Helped |
| Use only PASCAL view-consistency supervision rather than ShapeNet 3D supervision | Single-view reconstruction · PASCAL VOC chair | mean IoU: 0.33 -> 0.25 | Hurt |
| Use DRC rather than depth fusion with clean depth observations | ShapeNet reconstruction · ShapeNet airplane | mean IoU: 0.54 -> 0.49 | Hurt |
Closest papers by meaning
- Learning View Priors for Single-View 3D Reconstruction (CVPR 2019 · similarity 0.88)
- ConsistNet: Enforcing 3D Consistency for Multi-view Images Diffusion (CVPR 2024 · similarity 0.87)
- Fostering Generalization in Single-View 3D Reconstruction by Learning a Hierarchy of Local and Global Shape Priors (CVPR 2021 · similarity 0.87)
- Learning a Multi-View Stereo Machine (NeurIPS 2017 · similarity 0.87)
- Shape-Pose Ambiguity in Learning 3D Reconstruction from Images (AAAI 2021 · similarity 0.87)
- MVD-Fusion: Single-view 3D via Depth-consistent Multi-view Generation (CVPR 2024 · similarity 0.85)
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.