Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Light Field Super-Resolution With Zero-Shot Learning
Check 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 |
|---|---|---|---|
| Use input-only zero-shot learning with pre-upsampling, view alignment, and multi-view aggregation | Light field SR · Stan, EPFL, HCI1, and HCI2, scale 2 | PSNR (dB): 37.91 vs 37.03; 34.51 vs 34.00; 39.74 vs 38.86; 37.75 vs 37.08 | Helped |
| Use zero-shot SR rather than the strongest reported source-trained deep baseline | Light field SR · Stan, EPFL, HCI1, and HCI2, scale 2 | PSNR (dB): 37.91 vs 37.92; 34.51 vs 34.45; 39.74 vs 39.60; 37.75 vs 37.82 | Mixed |
| Apply error-guided fine-tuning to the source-pretrained model | Light field SR · Stan, EPFL, HCI1, and HCI2, scale 2, SAE source | PSNR (dB): 37.36 -> 37.97; 34.29 -> 34.64; 39.42 -> 39.98; 37.56 -> 37.85 | Helped |
| Apply error-guided fine-tuning to the source-pretrained model | Light field SR · Stan, EPFL, HCI1, and HCI2, scale 2, HFUT source | PSNR (dB): 37.55 -> 38.27; 34.55 -> 35.21; 39.42 -> 40.38; 37.22 -> 38.22 | Helped |
| Use error-guided rather than plain fine-tuning (ablation) | Ours-FT · Stanford and HCI2, scale 2 | PSNR (dB): about 0.4 dB | Helped |
| Divide SR into sub-tasks instead of using an end-to-end network (ablation) | Zero-shot light field SR · Stanford | PSNR (dB): 2.75/1.81/1.97 dB | Helped |
| Replace VDSR pre-upsampling with bicubic interpolation (ablation) | Zero-shot light field SR · Stanford and HCI2, scale 2 | PSNR (dB) | Hurt |
| Remove view alignment before aggregation (ablation) | Zero-shot light field SR · Stanford and HCI2, scale 2 | PSNR (dB): 0.73/0.62 dB | Hurt |
| Remove the back-projection loss (ablation) | AggreNet in zero-shot light field SR · Stanford and HCI2, scale 2 | PSNR (dB): 0.22/0.25 dB | Hurt |
| Remove the joint training stage (ablation) | Zero-shot light field SR · Stanford and HCI2, scale 2 | PSNR (dB): about 0.2 dB | Hurt |
Closest papers by meaning
- Learning Light Field Angular Super-Resolution via a Geometry-Aware Network (AAAI 2020 · similarity 0.90)
- Detail-Preserving Transformer for Light Field Image Super-resolution (AAAI 2022 · similarity 0.90)
- CutMIB: Boosting Light Field Super-Resolution via Multi-View Image Blending (CVPR 2023 · similarity 0.89)
- NeLF-Pro: Neural Light Field Probes for Multi-Scale Novel View Synthesis (CVPR 2024 · similarity 0.87)
- ZeroRF: Fast Sparse View 360deg Reconstruction with Zero Pretraining (CVPR 2024 · similarity 0.86)
- Occlusion-Embedded Hybrid Transformer for Light Field Super-Resolution (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.