Research map / 3D vision / Depth and stereo
Attention-Based View Selection Networks for Light-Field Disparity Estimation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 attention-based view selection (free attention) (ablation) | LFattNet · 4D LF ablation | MSE*100: 1.438 -> 1.284 | Helped |
| constrain attention map to be symmetric (ablation) | LFattNet free attention · 4D LF ablation | MSE*100: 1.284 -> 1.174 | Helped |
| constrain attention map to be radially symmetric (ablation) | LFattNet symmetric attention · 4D LF ablation | MSE*100: 1.174 -> 0.982 | Helped |
| radial attention view selection vs pre-selected views (Epinet) at similar parameters | light field disparity · 4D LF ablation | Badpix 0.07: 3.91 -> 2.45 | Helped |
| LFattNet vs Epinet-fcn | light field disparity · 4D LF Test Bedroom | Badpix 0.01: 33.99 -> 13.33 | Helped |
| LFattNet vs Epinet-fcn | light field disparity · 4D LF Test Bedroom | MSE*100: 0.213 -> 0.366 | Hurt |
| LFattNet vs Alperovich et al. | light field disparity · Alperovich dataset | MSE*100: 18.559 -> 2.866 | Helped |
Closest papers by meaning
- Attention-based Multi-Level Fusion Network for Light Field Depth Estimation (AAAI 2021 · similarity 0.91)
- Towards Multimodal Depth Estimation From Light Fields (CVPR 2022 · similarity 0.88)
- EPINET: A Fully-Convolutional Neural Network Using Epipolar Geometry for Depth From Light Field Images (CVPR 2018 · similarity 0.88)
- Learning Fused Pixel and Feature-Based View Reconstructions for Light Fields (CVPR 2020 · similarity 0.88)
- Occlusion-Aware Cost Constructor for Light Field Depth Estimation (CVPR 2022 · similarity 0.87)
- Convolutional Networks for Shape From Light Field (CVPR 2016 · similarity 0.87)
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.