/forgery-recognition

Handwriting style extraction and emulation

Primary LanguageJupyter Notebook

Forgery Recognition

The project is broken into three parts, each part run individually.

  1. GAN: Follow the Baseline Gan letter generator code jupyter notebook or run the notebook step-by-step
  2. Siamese Network: Follow the Siamese network jupyter notebook
  3. Convolutional GAN: Refer to the README within the CNN_GAN folder

Content of other folders:

  • generated_images: GAN and CNN_GAN generated images, alongside the original EMNIST words
  • siamese: forgery data for the Siamese model
  • letter_recognition: deprecated, not used
  • scribe: examples generated from Graves' LSTM implementation