artNet Genre Classifier

Inspiration

wikiart.org has many paintings of different genres and there is a need to make better interfaces to order, sort and search them

What it does

This is a classifier, which accurately classifies the given painting into 10 genres including landscape, portrait etc

How I built it

Built it using the Keras library with Hyperparameter optimisation done in Hyperas and Elephas library. Distributed optimization is performed using Spark nodes.

Challenges I ran into

Setting up Spark properly was a big challenge faced

What I learned

Learned quite a lot about Convolutional Neural Network architectures and their effectiveness.

What's next for artNet Genre Classifier

Next would be to do Painter recognition as well as Painting style classification

Built With

• keras
• python
• tensorflow