Research map / 3D vision / Depth and stereo
Generalizable Non-Line-of-Sight Imaging with Learnable Physical Priors
Shida Sun, Yue Li, Yueyi Zhang, Zhiwei Xiong
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 Learnable Path Compensation (LPC) (ablation) | LFE-based network without LPC and APF · public real-world data (10 min acquisition) | visual detail | Improved |
| add Adaptive Phasor Field (APF) (ablation) | LFE-based network without LPC and APF · public real-world data (10 min and 0.3 min) | background artifacts | Improved |
| add both LPC and APF (ablation) | LFE-based network baseline · public real-world data | visual quality | Improved |
| replace LFE with proposed LPC+APF network | LFE · Seen synthetic test set | PSNR: 23.27 -> 23.99 | Improved |
| compare with best traditional approach RSD | Ours · Seen synthetic test set | PSNR: 2.25 dB increase in PSNR | Improved |
| compare with I-K and NLOST | Ours · Seen synthetic test set | PSNR: 0.55 dB and 0.25 dB improvement in PSNR | Improved |
| compare with NLOST on depth | Ours · Seen synthetic test set | RMSE and MAD: 3.10% and 8.77% | Improved |
| compare with transformer NLOST on memory | Ours · Seen synthetic test set | GPU memory: 38 GB -> 17 GB | Improved |
| lower SNR to 3 dB on Unseen scenes | Ours · Unseen test set | SSIM: 0.8577 -> 0.8645 | No clear effect |
| compare with NLOST at 5 dB depth | Ours · Unseen test set (5 dB SNR) | depth RMSE: 0.0909 vs 0.0914 | Hurt |
| compare with NLOST at 10 dB SSIM | Ours · Unseen test set (10 dB SNR) | SSIM: 0.7747 -> 0.8577 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- NLOST — NLOST: Non-Line-of-Sight Imaging With Transformer (CVPR 2023) · better on 14 of 14 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Non-Line-of-Sight Imaging With Signal Superresolution Network (CVPR 2023 · similarity 0.89)
- Non-line-of-Sight Imaging via Neural Transient Fields (TPAMI 2021 · similarity 0.89)
- Enhancing Non-line-of-sight Imaging via Learnable Inverse Kernel and Attention Mechanisms (ICCV 2023 · similarity 0.88)
- Learning to Enhance Aperture Phasor Field for Non-Line-of-Sight Imaging (ECCV 2024 · similarity 0.88)
- Virtual Scanning: Unsupervised Non-line-of-sight Imaging from Irregularly Undersampled Transients (NeurIPS 2024 · similarity 0.88)
- WaveMo: Learning Wavefront Modulations to See Through Scattering (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.
