Exclusively Dark (ExDARK) dataset which to the best of our knowledge, is the largest collection of low-light images taken in very low-light environments to twilight (i.e 10 different conditions) to-date with image class and object level annotations.
Thanks for collecting and publishing the ExDark dataset.
I'm interested in some more details about the ResNet-50 classification experiments you performed in the accompanying paper, Getting to Know Low-light Images with The Exclusively Dark Dataset (https://arxiv.org/pdf/1805.11227.pdf). As MS COCO (and to some extent ExDark) images generally contain multiple objects, I was wondering how you used the dataset(s) for a classification task. Did you crop the image to show only the object of the corresponding class? Could you maybe share the MS COCO subset you used?
Thanks for collecting and publishing the ExDark dataset. I'm interested in some more details about the ResNet-50 classification experiments you performed in the accompanying paper,
Getting to Know Low-light Images with The Exclusively Dark Dataset
(https://arxiv.org/pdf/1805.11227.pdf). As MS COCO (and to some extent ExDark) images generally contain multiple objects, I was wondering how you used the dataset(s) for a classification task. Did you crop the image to show only the object of the corresponding class? Could you maybe share the MS COCO subset you used?