Research map / Learning theory and algorithms / Matrix and tensor methods
Convolutional Approximations to the General Non-Line-of-Sight Imaging Operator
Byeongjoo Ahn, Akshat Dave, Ashok Veeraraghavan, Ioannis Gkioulekas, Aswin C. Sankaranarayanan
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 |
|---|---|---|---|
| solve the Gram-operator deconvolution instead of full linear reconstruction | NLOS reconstruction (S-shape, 5D transient) · Simulated S-shape scene | runtime: 70x faster with similar reconstruction quality | Improved |
| use exact inverse kernel deconvolution instead of Laplacian filtered backprojection | Ours (w/o priors) · Simulated Stanford bunny | mean absolute depth error (cm): 1.01 cm -> 0.99 cm | Improved |
| use exact inverse kernel deconvolution instead of Laplacian filtered backprojection | Ours (w/o priors) · Simulated Stanford bunny | root mean square depth error (cm): 2.38 cm -> 2.33 cm | Improved |
| use Gram-based method at 0.24 cm voxels instead of LCT at 0.78 cm voxels | Ours (w/o priors) · Simulated Stanford bunny | mean absolute depth error (cm): 1.12 cm -> 0.99 cm | Improved |
| add positivity, total variation and sparsity priors (ablation) | Gram-based NLOS reconstruction · Real confocal transient | reconstruction quality (qualitative) | Improved |
| use the Gram method without priors instead of filtered backprojection | Gram-based NLOS reconstruction · Real confocal transient | reconstruction quality (qualitative) | Improved |
| use Gram-based method instead of LCT at low scanning resolution | Gram-based NLOS reconstruction · Rendered confocal transient, 64x64 scan on 0.5 m wall | lateral detail (qualitative) | Improved |
Closest papers by meaning
- Domain Reduction Strategy for Non-Line-of-Sight Imaging (ECCV 2024 · similarity 0.86)
- Generalizable Non-Line-of-Sight Imaging with Learnable Physical Priors (ICCV 2025 · similarity 0.86)
- Non-Line-of-Sight Imaging With Signal Superresolution Network (CVPR 2023 · similarity 0.85)
- Enhancing Non-line-of-sight Imaging via Learnable Inverse Kernel and Attention Mechanisms (ICCV 2023 · similarity 0.85)
- Curvature Regularization for Non-Line-of-Sight Imaging From Under-Sampled Data (TPAMI 2024 · similarity 0.85)
- Virtual Scanning: Unsupervised Non-line-of-sight Imaging from Irregularly Undersampled Transients (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.
