The premise is simple enough. You are given a word, and have twenty seconds to draw it well enough for the app to figure out what you are drawing. There are six different pictures to draw and the aim is to have the app recognise as many as possible.
It has resulted in a lot of fun today and we thought we would share some of our efforts! Let’s see how well we did.
So I must hold my hands up, drawing was never a strong point for me. I did feel hard done by with my lobster not being recognised though:
Fortunately my colleagues are a lot more talented! This is how to get a neural network to recognise a lobster:
A great effort here with a close to full marks attempt:
This is how it’s done:
And finally, we thought that ‘Draw animal migration’ in twenty seconds was a tall order and deserved a special mention! Here was our best attempt which, sadly, did not get recognised:
So overall, we have very much enjoyed experimenting with this web app today. It is great to see a creative and fun way to employ pattern matching techniques and neural networks in technology, and it is certainly exciting to see how things progress in the machine learning area of technology moving forward.