Megmondja neked egy keprol, mennyire ertekelnek az emberek
=======================
Research on inferring and rating photo aesthetics automatically using visual features. Computer program provides unbiased assessment of photos.
Professional art critics endlessly debate the merits of various photographs -- heck, that's the entire premise of a photography contest! -- but what if a computer could reliably do the same thing? Researchers at Penn State put together a photo ranking algorithm that you can now test for yourself on the web. It's called Acquine, and it's based on data from large numbers of human ratings that determined what people find aesthetically pleasing in a photo.
For best results, give Acquine a high-resolution color photo, at least 600x600 pixels. It will give the photo an unbiased rating between 0 and 100. There's a library of the photos people have already uploaded, and you check them out in rank order to see if the algorithm gives the same results you would have. It's definitely an interesting exercise, and a thought-provoking comment on the challenges of replicating human aesthetic judgments through software.