Research map / Video / Video segmentation and tracking
SeLFVi: Self-Supervised Light-Field Video Reconstruction From Stereo Video
Prasan Shedligeri, Florian Schiffers, Sushobhan Ghosh, Oliver Cossairt, Kaushik Mitra
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 epipolar geometric consistency loss L_geo (ablation) | SeLFVi · LF video ablation | PSNR: 19.20 -> 32.39 | Improved |
| add stereo photometric loss L_stereo (ablation) | SeLFVi · LF video ablation | PSNR: 32.20 -> 32.39 | No clear effect |
| add temporal loss L_temp (ablation) | SeLFVi · LF video ablation | PSNR: 31.98 -> 32.39 | No clear effect |
| add tensor-display (low-rank) layer to the model (ablation) | SeLFVi · LF video ablation | PSNR: 30.50 -> 32.39 | Improved |
| add tensor-display layer when L_geo is absent (ablation) | SeLFVi · LF video ablation | PSNR: 6.04 -> 19.20 | Improved |
| increase rank M of F from 1 to 12 (ablation) | SeLFVi · 7x7 LF output | PSNR: 31.43 -> 32.39 | Improved |
| increase rank M of F from 1 to 12 (ablation) | SeLFVi · 7x7 LF output | inference time: 0.103 -> 0.381 | Hurt |
| use direct regression instead of low-rank intermediate representation (ablation) | SeLFVi · 7x7 LF output | inference time: 0.381 -> 0.108 | Improved |
| use SeLFVi instead of X-fields 2-view | LF video from stereo · Average of three datasets | PSNR: 27.12 -> 33.29 | Improved |
| use SeLFVi instead of X-fields 4-view (which needs four views) | LF video from stereo · Average of three datasets | PSNR: 30.84 -> 33.29 | Improved |
| use SeLFVi instead of HighRes warping | LF video from stereo · Average of three datasets | LPIPS: 0.063 -> 0.071 | Hurt |
| use SeLFVi instead of X-fields 4-view | LF video from stereo · Temporal consistency | warping MAE: 1.73 -> 1.58 | Improved |
Closest papers by meaning
- Synthesizing Light Field Video from Monocular Video (ECCV 2022 · similarity 0.94)
- Fast Light-Field Disparity Estimation With Multi-Disparity-Scale Cost Aggregation (ICCV 2021 · similarity 0.86)
- Light Field Reconstruction via Deep Adaptive Fusion of Hybrid Lenses (TPAMI 2023 · similarity 0.86)
- Learning Fused Pixel and Feature-Based View Reconstructions for Light Fields (CVPR 2020 · similarity 0.86)
- Event Fields: Capturing Light Fields at High Speed, Resolution, and Dynamic Range (CVPR 2025 · similarity 0.85)
- S2D-LFE: Sparse-to-Dense Light Field Event Generation (CVPR 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.
