Table 1 Comparison of test sets used for evaluation of performance and comparative analysis.
From: A multi scale spatial attention based zero shot learning framework for low light image enhancement
Dataset | #Images | Paired | Real/synthetic | Description |
|---|---|---|---|---|
DarkBDD54 | 100 | No | Real | 100 lowlight images selected from the orignal BDD10K76 and created a new test set called as DarkBDD54 |
DarkCityScape54 | 150 | No | Real | Gamma correction applied on selected 150 images from original CityScape77 dataset to create DarkCityScape54 |
DICM78 | 69 | No | Real | Diverse indoor and outdoor low-light images captured with commercial digital cameras |
LIME79 | 10 | No | Real | Real-world low-light images |
LOL80 | 15 | Yes | Real/synthetic | Paired widely used benchmark for supervised enhancement methods. Selected 15 images |
LOL-v281 | 100+100 | Yes | Real | Improved version with more diverse scenes and better alignment; used 100 testing images for both real and synthetic |
MEF82 | 17 | No | Real | Multi-exposure fusion sequences |
NPE83 | 8 | No | Real | Nighttime photographs with varied exposure levels |
VV84 | 24 | No | Real | Night road scenarios; used in autonomous driving and low-light enhancement studies |
