Is there a way/algorithm to train a model to enhance low light images without actually having a paired dataset of normal, low-light? Coz there are very limited datasets like that and I feel this limited the generalization to real world images.
Is it possible to increase the sample count of the low light images?
1 Like