Research map / 3D vision / Depth and stereo
LoLep: Single-View View Synthesis with Locally-Learned Planes and Self-Attention Occlusion Inference
Cong Wang, Yu-Ping Wang, Dinesh Manocha
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 |
|---|---|---|---|
| learn plane locations locally with the disparity sampler (U-opt) instead of random locations (ablation) | LoLep-16 · KITTI | RV: 839.17 -> 182.13 | Improved |
| learn plane locations globally instead of locally (ablation) | LoLep-16 · KITTI | PSNR: 21.48 -> 19.64 | Hurt |
| divide disparity space equally instead of random sampling in bins (ablation) | LoLep-16 · KITTI | PSNR: 21.48 -> 21.28 | Hurt |
| add occlusion-aware reprojection loss (ablation) | LoLep-16 (random locations) · KITTI | PSNR: 21.48 -> 21.70 | Improved |
| reprojection loss without the occlusion mask (ablation) | LoLep-16 (random locations) · KITTI | PSNR: 21.48 -> 21.32 | Hurt |
| add BS-SA module (M=400) to random-location model (ablation) | LoLep-16 · KITTI | PSNR: 21.48 -> 21.62 | Improved |
| increase BS-SA sampled queries M from 100 to 1500 (ablation) | LoLep-16 · KITTI | PSNR: 21.57 -> 21.94 | Improved |
| too small M (100) in BS-SA (ablation) | LoLep-16 · KITTI | PSNR: 21.74 -> 21.57 | Hurt |
| replace original self-attention with BS-SA (ablation) | LoLep-16 · KITTI | PSNR: 21.78 -> 21.76 | No clear effect |
| use LoLep-32 instead of MINE-64 (fewer planes) | LoLep · KITTI | LPIPS: 0.127 -> 0.122 | Improved |
| replace MINE with LoLep-64 for depth | 64-plane models trained on RealEstate10K · NYU-Depth V2 | RV: 3.41 -> 2.53 | Improved |
Closest papers by meaning
- Single-View View Synthesis With Multiplane Images (CVPR 2020 · similarity 0.86)
- DeOcc-1-to-3: 3D De-Occlusion from a Single Image via Self-Supervised Multi-View Diffusion (AAAI 2026 · similarity 0.85)
- LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias (ICLR 2025 · similarity 0.84)
- Novel View Synthesis with View-Dependent Effects from a Single Image (CVPR 2024 · similarity 0.84)
- Novel View Synthesis with Pixel-Space Diffusion Models (CVPR 2025 · similarity 0.84)
- Less Decoder is More Encoder: Geometric Representation Learning from Novel View Synthesis (NeurIPS 2026 · 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.
