Describe

•Input: Underexposed photo. 

•Output: Full-res enhanced image. 

•Dataset: MIT-Adobe FiveK ,a new dataset of 3,000 underexposed image pairs. 

•Framework: TensorFlow 

•Configuration: NVidia Titan X Pascal GPU

Major Contributions

•We propose a network for enhancing underexposed photos by estimating an image-to-illumination map- ping, and design a new loss function based on various illumination constraints and priors. 

•We prepare a new dataset of 3,000 underexposed images, each with an expert-retouched reference. 

•We perform evaluation on our method using existing and new datasets, and demonstrate the superiority of our method qualitatively and quantitatively.