Benchmark for image or photo enhancement algorithm

Question

I was wondering if anybody knows approaches or available set of images used to benchmark image enhancement algorithms. The type of image enhancement meant here is improving the aesthetic look of an image by changing its exposure, brightness, white-balance, noise reduction, sharpening etc.

A desired benchmark would be to have a collection of images. Each image has two version - original and enhanced. Then the goal of the benchmark would be to compare the output of the image enhancing algorithm with he enhanced version of the image from the image collection.

I would appreciate your feedback and pointers.

Also, I would appreciate if you would be able to comment on the possible metrics to use during the benchmarking.

Answer 1

Try the MIT-Adobe FiveK Dataset: http://groups.csail.mit.edu/graphics/fivek_dataset/