<- Back to Datasets

REDS

REalistic and Dynamic Scenes dataset

REDS

We propose the REDS dataset, a novel REalistic and Dynamic Scenes dataset of 720 × 1280 resolution for training and benchmarking example-based video deblurring and super-resolution. REDS is intended to complement the existing video deblurring and SR datasets to increase the content diversity and provide more realism in degradation, especially, motion blur.

View this Dataset
->
SNU, Seoul, Korea
View author website
Task
Image Enhancement
Annotation Types
Items
Classes
Labels
Models using this dataset
Last updated on 
January 20, 2022
Licensed under 
Unknown
Gain control of your training data
15,000+ ML engineers can’t be wrong