Hi! Quite original as a concept. The idea to use ML to drive a game is a fun idea.
But it's terribly difficult for the doodles to be recognized. Either simple forms or complex ones.
After multiple tries, I finally understood that the sun needed rays to be recognized as such. I've only been able to create a whale (while trying to draw a simple fish). I was stunned to see the game suggest me to draw a turtle and other complex animals. How? I'm not even able to draw a mountain.
The concept is great on paper, but is difficult to implement well. That makes the game not fun to play, unfortunately.
It shows that a lot of work have been put in this project. It's even more unfortunate that we cannot see the fruits of your labour as everything is dictated by the ability of Tensorflow to recognise my doodles.
Keep it up, it was a nice try!
