Research map / Video / Video segmentation and tracking
Detail-Revealing Deep Video Super-Resolution
Xin Tao, Hongyun Gao, Renjie Liao, Jue Wang, Jiaya Jia
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 SPMC layer (sub-pixel motion compensation) in place of backward warping plus bicubic (ablation) | Ours F3 · SPMCS x4 | PSNR: 29.23 -> 29.69 | Improved |
| add bicubic upsampled reference frame as extra DF-net input (ablation) | Ours F3 · SPMCS x4 | PSNR: 29.69 -> 29.67 | No clear effect |
| add zero-upsampled reference frame as extra DF-net input (ablation) | Ours F3 · SPMCS x4 | PSNR: 29.69 -> 29.65 | No clear effect |
| feed 3 copies of the same reference frame instead of 3 consecutive frames (ablation) | Ours F3 · qualitative examples | visual detail | Hurt |
| increase input frames from 3 to 5 | Ours · SPMCS x4 | PSNR: 29.69 -> 29.89 | Improved |
| increase input frames from 3 to 5 | Ours · SPMCS x2 | PSNR: 36.71 -> 36.62 | No clear effect |
| increase input frames from 3 to 5 | Ours · SPMCS x3 | PSNR: 31.92 -> 32.10 | Improved |
| use 3 frames instead of 1 frame | Ours · SPMCS x4 | PSNR: 28.80 -> 29.69 | Improved |
| use 3 frames instead of 1 frame | Ours · SPMCS x2 | PSNR: 36.23 -> 36.71 | Improved |
| use multi-frame SPMC network (F3) instead of VSRnet | Ours · SPMCS x4 | PSNR: 24.76 -> 29.69 | Improved |
| use single-frame version of Ours instead of VDSR | Ours F1 · Set5 x4 | PSNR: 31.35 -> 30.96 | Hurt |
| use Ours F1 instead of VDSR | Ours F1 · SPMCS x2 | PSNR: 36.14 -> 36.23 | No clear effect |
Closest papers by meaning
- Image Super-Resolution Using Deep Convolutional Networks (TPAMI 2016 · similarity 0.89)
- Real-Time Video Super-Resolution With Spatio-Temporal Networks and Motion Compensation (CVPR 2017 · similarity 0.89)
- Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network (CVPR 2016 · similarity 0.88)
- Video Face Super-Resolution with Motion-Adaptive Feedback Cell (AAAI 2020 · similarity 0.88)
- Building an End-to-End Spatial-Temporal Convolutional Network for Video Super-Resolution (AAAI 2017 · similarity 0.88)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · 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.
