Research map / 3D vision / Depth and stereo
Photon-Efficient 3D Imaging with A Non-Local Neural Network
Jiayong Peng, Zhiwei Xiong, Xin Huang, Zheng-Ping Li, Dong Liu, Feihu Xu
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 the non-local block over the plain backbone (ablation) | dense dilated fusion backbone · simulated, noise 2:50 | RMSE: 49% improvement | Improved |
| add the non-local block over the plain backbone (ablation) | dense dilated fusion backbone · simulated, noise 10:2 | RMSE: 42% improvement | Improved |
| add the non-local block to a baseline with downsampling (BAS) (ablation) | BAS · simulated, noise 2:50 | RMSE: 18% improvement | Improved |
| add the non-local block to a baseline with downsampling (BAS) (ablation) | BAS · simulated, noise 10:2 | RMSE: 6% improvement | Improved |
| add the non-local block for generalization to unseen noise levels (ablation) | BAS vs NLB · 3 extremely low SBR test cases | RMSE: BAS degrades 60%, NLB drops 24% | Improved |
| downsample temporal dimension of features from 1024 to 64 channels (ablation) | non-local network · simulated | RMSE: turning point at 64 channels | Mixed |
| use this network instead of Rapp | photon-efficient depth reconstruction · Middlebury simulated, 9 typical noise levels | RMSE: 0.0385 -> 0.0184 | Improved |
| use this network instead of Rapp on extremely low photon/SBR levels | photon-efficient depth reconstruction · Middlebury simulated, 3 low-end noise levels | RMSE: 0.2794 -> 0.0490 | Improved |
| use this network instead of Lindell | photon-efficient depth reconstruction · Running time | Time (s): 2.91 -> 0.55 | Improved |
Closest papers by meaning
- Boosting Photon-Efficient Image Reconstruction with A Unified Deep Neural Network (TPAMI 2022 · similarity 0.93)
- Single-Photon Imaging in Complex Scenarios via Physics-Informed Deep Neural Networks (TPAMI 2026 · similarity 0.89)
- Towards 3D Vision with Low-Cost Single-Photon Cameras (CVPR 2024 · similarity 0.88)
- Learned Compressive Representations for Single-Photon 3D Imaging (ICCV 2023 · similarity 0.87)
- Count-Free Single-Photon 3D Imaging With Race Logic (TPAMI 2023 · similarity 0.86)
- Separating the Wheat from the Chaff: Spatio-Temporal Transformer with View-interweaved Attention for Photon-Efficient Depth Sensing (AAAI 2025 · 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.
