What makes a good photo? To some it could be the technicality of the shot, where the subject is perfectly exposed, edges are sharp where they need to be, and white balance is accurate. To others a good photo could be about the composition, the juxtaposition of the subject on the background, the use of contrasting colors, and so on.
That being said if even humans can’t necessarily always agree with what makes a good photo, can AI do a better job? That’s something that Google wants to try and figure out, by using convolution neural networks to help AI identify photos that we humans rate as good or attractive.
According to Google, “Our proposed network can be used to not only score images reliably and with high correlation to human perception, but also it is useful for a variety of labor intensive and subjective tasks such as intelligent photo editing, optimizing visual quality for increased user engagement, or minimizing perceived visual errors in an imaging pipeline.”
Google isn’t alone in trying to teach AI what humans might deem as attractive. Recently Adobe has pushed out an update to Lightroom in which it uses AI to help edit photos to a level that it thinks is aesthetically pleasing.