A company called EVERYPIXEL is building a stock image search engine to search across hundreds of millions of stock images.  They are also building a heuristic tool to rate images so that the can weed-out the bad ones.  Currently, the beta version of the heuristic tool is available to use for free.  It can’t really tell you if an image is truly great because it’s too primitive at this point.  But no matter how complex you think aesthetics is, remember IBM Watson https://en.wikipedia.org/wiki/Watson_(computer) is just a stepping stone for things to come. 
I’ve played with it and it clearly has sense of what appeals to the human eye.  Here’s a link to the tool:  https://everypixel.com/aesthetics
You click on the camera icon in the square, and it will produce a file selection window.  Select your file (an 800-pixel-wide or high image works as well as a larger image, but is evaluated in a few seconds, rather than minutes, so use small images).  The tool will return a level of confidence that your image is “awesome”.

99.2% seems high, but I have at least a couple images that score 100.0%, and I don't believe that they are more aesthetically pleasing.  I also have images that I like they score 2-3%.  The tool can be used to evaluate the effect of adjustments that you make in Photoshop, if you have ever wondered how much you have improved an image’s aesthetics  after working on it for an hour. LOL!  I’m sure you will find uses.
Here’s the instructions to this game:
1.    Post a link to one of your photos with the score you received printed below it.
2.    The score must be either a.) greater than 20%, or b.) the score of your highest scoring image.
3.    Remember, it isn’t a perfect tool.  Also, aesthetics is a good thing, but it isn't everything.
According to a credible post in another thread, "commercial"-type images of people can score very high. I tested one and got 100%.  I tested another and got 3.8%

Example:

89.7%

If your score is not very high, don't worry.  It has some bugs.  We could discuss ideas of what might produce a higher score if you ask.  Also, we need to determine if our results are consistent between out respective machines.