Research map / 3D vision / Depth and stereo
Recovering Parametric Scenes from Very Few Time-of-Flight Pixels
Carter Sifferman, Yiquan Li, Yiming Li, Fangzhou Mu, Michael Gleicher, Mohit Gupta, Yin Li
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 differentiable-rendering pose refiner to feedforward network (ablation) | Ours: Feedforward · 3D-printed objects (Mean) | AUC-ADD: 70.96 -> 80.71 | Improved |
| add pose refiner to feedforward network (ablation) | Ours: Feedforward · YCB symmetric objects (Mean) | AUC-ADD-S: 91.83 -> 92.20 | Improved |
| add pose refiner to feedforward network (ablation) | Ours: Feedforward · YCB SPAM | AUC-ADD-S: 90.04 -> 90.00 | No clear effect |
| use diffuse ToF sensors with learned method instead of 1-pixel point cloud with oracle init (ablation) | 1 Px Point Cloud · 3D-printed objects (Mean) | AUC-ADD: 59.73 -> 80.71 | Improved |
| fine-tune feedforward on real data (ablation) | Fully-sim trained model · "2" object | AUC-ADD (FF + Refiner): 83.47 -> 90.36 | Improved |
| use pretraining on sim then fine-tuning on real for hand pose (ablation) | Real-only training · Hand pose | PA-MPJPE (mm): 9.98 -> 8.18 | Improved |
| use sim-only training instead of real for hand pose (ablation) | Real-only training · Hand pose | PA-MPJPE (mm): 9.98 -> 19.56 | Hurt |
| change tabletop from matte white to matte black (ablation) | FF + Refiner · "2" object | AUC-ADD: 83.47 -> 79.55 | Hurt |
| change object to spotted black and white (ablation) | FF + Refiner · "2" object | AUC-ADD: 83.47 -> 61.49 | Hurt |
| increase ambient light to 3000 lux (ablation) | FF + Refiner · "2" object | AUC-ADD: 72.45 -> 25.45 | Hurt |
| replace jitter kernel with Dirac delta (remove accurate jitter kernel) (ablation) | FF + Refiner · "2" object | AUC-ADD: 83.47 -> 24.50 | Hurt |
| use incorrect temporal bin size (~10% smaller) (ablation) | FF + Refiner · "2" object | AUC-ADD: 83.47 -> 33.23 | Hurt |
Closest papers by meaning
- 3D Scene Inference from Transient Histograms (ECCV 2022 · similarity 0.87)
- Time of the Flight of the Gaussians: Optimizing Depth Indirectly in Dynamic Radiance Fields (CVPR 2025 · similarity 0.87)
- Towards 3D Vision with Low-Cost Single-Photon Cameras (CVPR 2024 · similarity 0.86)
- Tackling 3D ToF Artifacts Through Learning and the FLAT Dataset (ECCV 2018 · similarity 0.86)
- PlatoNeRF: 3D Reconstruction in Plato's Cave via Single-View Two-Bounce Lidar (CVPR 2024 · similarity 0.86)
- Snapshot Lidar: Fourier Embedding of Amplitude and Phase for Single-Image Depth Reconstruction (CVPR 2024 · 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.
