Skip to content

Latest commit

 

History

History
14 lines (10 loc) · 505 Bytes

README.md

File metadata and controls

14 lines (10 loc) · 505 Bytes

Artistic Style Recognition

Attempts to guess a painting's artistic style using a deep neural network. Uses transfer learning on an optimized convolutional SqueezeNet applied to data (not included) from the WikiArt database.

Built for a Machine Learning course at Oxford University.

Accuracy vs. Epoch Loss vs. Epoch

TODO:

  • Add ability to use multiple architectures, not just SqueezeNet
  • Add CUDNN benchmarking
  • Add ability to properly disable CUDA (try import?)