Quick Draw

Inspiration

We thought it would be fun to experiment with computer vision in the context of art and doodles. Both of us enjoy art and we wanted to combine our interests with coding. Our work is inspired by Google's Quick Draw game.

What it does

Our model identifies what animal is being depicted in a drawing from a set of 10 options.

How we built it

We used a CNN and implemented it into a webapp.

Challenges we ran into

Initially, we tried to use an RNN, but we were having difficulty so we switched to a CNN and this resulted in a more successful and accurate model.

Accomplishments that we're proud of

Test accuracy of over 83%!

What we learned

This was our first time learning with neural networks, so we gained useful experience in terms of how to understand them and put them together.

What's next for Quick Draw

If we were to continue with this project, we would try to incorporate more than 10 categories of animal into the model in order to make it more complex. We might also want to implement a more sophisticated webapp.

Built With

• cnn
• python
• quickdraw