Research map / Point clouds / Point cloud classification
Revealing Scenes by Inverting Structure From Motion Reconstructions
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 |
|---|---|---|---|
| Add color to depth + SIFT input (ablation) | CoarseNet · MegaDepth and NYU test, 100% points | MAE: MegaDepth .221 -> .195; NYU .171 -> .093 | Helped |
| Add color to depth + SIFT input (ablation) | CoarseNet · MegaDepth and NYU test, 100% points | SSIM: MegaDepth .438 -> .448; NYU .539 -> .594 | Helped |
| Add SIFT descriptors to depth + color input (ablation) | CoarseNet · MegaDepth and NYU test, 100% points | MAE: MegaDepth .202 -> .195; NYU .111 -> .093 | Helped |
| Add SIFT descriptors to depth + color input (ablation) | CoarseNet · MegaDepth and NYU test, 100% points | SSIM: MegaDepth .403 -> .448; NYU .546 -> .594 | Helped |
| Add SIFT descriptors and color to depth-only input (ablation) | CoarseNet · MegaDepth and NYU test, 100% points | MAE: MegaDepth .253 -> .195; NYU .289 -> .093 | Helped |
| Add SIFT descriptors and color to depth-only input (ablation) | CoarseNet · MegaDepth and NYU test, 100% points | SSIM: MegaDepth .250 -> .448; NYU .197 -> .594 | Helped |
| Predict visibility and cull occluded points before reconstruction (ablation) | CoarseNet · MegaDepth and NYU, 100% points | SSIM: MegaDepth .445 -> .448; NYU .592 -> .595 | Helped |
| Predict visibility and cull occluded points before reconstruction (ablation) | CoarseNet · MegaDepth and NYU, 100% points | MAE: MegaDepth .195 -> .195; NYU .094 -> .092 | Mixed |
| Remove SIFT scale and orientation information (ablation) | Single-image SIFT inversion · Combined MegaDepth and NYU, 100% features | SSIM: .631 -> .578 | Hurt |
| Use descriptors from multiple unknown source images instead of one source (ablation) | Single-image SIFT inversion · Combined MegaDepth and NYU, 100% features | SSIM: .578 -> .524 | Hurt |
| Add color and SIFT attributes to visibility input (ablation) | VisibNet · MegaDepth and NYU test, 100% points | accuracy: MegaDepth .946 -> .950; NYU .908 -> .917 | Mixed |
Closest papers by meaning
- Efficient Privacy-Preserving Visual Localization Using 3D Ray Clouds (CVPR 2024 · similarity 0.89)
- ConcreTizer: Model Inversion Attack via Occupancy Classification and Dispersion Control for 3D Point Cloud Restoration (ICLR 2025 · similarity 0.86)
- Hide in Thicket: Generating Imperceptible and Rational Adversarial Perturbations on 3D Point Clouds (CVPR 2024 · similarity 0.84)
- Point2Real: Bridging the Gap between Point Cloud and Realistic Image for Open-World 3D Recognition (AAAI 2024 · similarity 0.84)
- Explicitly Perceiving and Preserving the Local Geometric Structures for 3D Point Cloud Attack (AAAI 2024 · similarity 0.84)
- GS-Hider: Hiding Messages into 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.84)
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.