Research map / 3D vision / Depth and stereo
Soft Shadow Diffusion (SSD): Physics-inspired Learning for 3D Computational Periscopy
Fadlullah A Raji, John Murray-Bruce
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 |
|---|---|---|---|
| optimize background, regularization constant and occluder jointly instead of assuming negligible background (ablation) | gradient-based voxel inversion · simulated and real NLOS data | convergence | Improved |
| add positivity constraint when background and noise are not negligible | gradient-based voxel inversion · real experiments | reconstruction quality | Improved |
| use learned SSD shape prior instead of gradient-based voxel optimization for complex occluders | 3D computational periscopy · complex occluder shapes | 3D reconstruction | Improved |
| increase ambient background illumination (lower signal-to-background ratio) | SSD · simulated and real NLOS data | 3D (Chamfer Distance) and 2D (MSE) reconstruction: 3D accurate at SBR around 10 dB; 2D degrades below 20 dB | Mixed |
| lower measurement signal-to-noise ratio to 10 dB | SSD · simulated data | 3D reconstruction quality | No clear effect |
| train SSD only on simulated ShapeNet data, test on real scenes | SSD · real NLOS experiments | qualitative reconstruction | Improved |
Closest papers by meaning
- Similarity-Consistent Likelihood Diffusion enables Hidden Person Detection from Wall Reflections (CVPR 2026 · similarity 0.88)
- Enhancing Non-line-of-sight Imaging via Learnable Inverse Kernel and Attention Mechanisms (ICCV 2023 · similarity 0.87)
- Deep Non-Line-of-Sight Reconstruction (CVPR 2020 · similarity 0.86)
- Virtual Scanning: Unsupervised Non-line-of-sight Imaging from Irregularly Undersampled Transients (NeurIPS 2024 · similarity 0.86)
- Coherent 3D Scene Diffusion From a Single RGB Image (NeurIPS 2024 · similarity 0.85)
- Diff-Shadow: Global-guided Diffusion Model for Shadow Removal (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.
