Multi-frame Super-resolution on BurstSR 1.0 (test)
3.42MORDeepJoint+RRDB
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DeepJoint+RRDBpipeline=Two-stage (demosaicking/denoising + super-resolution), weights=Pre-trained2021.01 | 3.42 | 8.9 | 42.13 | 0.088 | 0.957 | |
| Single Imagetraining=Joint denoising, demosiacking, and SR2021.01 | 2.41 | 18.3 | 44.02 | 0.051 | 0.972 | |
| HighRes-netdepth=Increased to match the depth of the proposed approach, training=Joint denoising, demosiacking, and SR2021.01 | 2.36 | 19.2 | 43.99 | 0.051 | 0.972 | |
| Deep Burst Super-Resolution2021.01 | 1.81 | 53.6 | 45.17 | 0.037 | 0.978 |