yinxusen / decipherment-images

unsupervised learning to decipher from historical manuscript images

Decipherment from Manuscript Images

This repo contains three parts of code:

• segmentation: segment character images out of a page.
• SNN: train an SNN model as the character images feature extractor.
• LM-GMM: decipher from transcriptions or image features.

cite our paper:

@article{DBLP:journals/corr/abs-1810-04297,
  author    = {Xusen Yin and
               Nada Aldarrab and
               Be{\'{a}}ta Megyesi and
               Kevin Knight},
  title     = {Decipherment of Historical Manuscript Images},
  journal   = {CoRR},
  volume    = {abs/1810.04297},
  year      = {2018},
  url       = {http://arxiv.org/abs/1810.04297},
  archivePrefix = {arXiv},
  eprint    = {1810.04297},
  timestamp = {Tue, 30 Oct 2018 20:39:56 +0100},
  biburl    = {https://dblp.org/rec/bib/journals/corr/abs-1810-04297},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}