/VistaOCR

ISI's Optical Character Recognition (OCR) software for machine-print and handwriting data

Primary LanguagePythonApache License 2.0Apache-2.0

VistaOCR

ISI's Optical Character Recognition (OCR) software for machine-print and handwriting data

Publications

"How to Efficiently Increase Resolution in Neural OCR Models". Stephen Rawls, Huaigu Cao, Joe Mathai, Prem Natarajan. IEEE Workshop on Arabic Script Analysis and Recognition (ASAR) 2018.

"Combining Convolutional Neural Networks and LSTMs for Segmentation Free OCR". Stephen Rawls, Huaigu Cao, Senthil Kumar, Prem Natarajan. International Conference on Document Analysis and Recognition (ICDAR) 2017.

"Combining Deep Learning and Language Modeling for Segmentation-free OCR From Raw Pixels". Stephen Rawls, Huaigu Cao, Ekraam Sabir, Prem Natarajan. IEEE Workshop on Arabic Script Analysis and Recognition (ASAR) 2017.

Model

VistaOCR Model Diagram

Pretrained Models

Coming Soon. Pre-trained models for English, French, and Arabic Handwriting

Performance Numbers

Coming soon. Expected character and word error rates from public datasets.

How to Train

Coming soon.

How to Decode using Existing Model

Coming soon.

Citation

@inproceedings{vistaocr,
  author    = {Stephen Rawls and Huaigu Cao and Senthil Kumar and Prem Natarjan},
  title     = {Combining Convolutional Neural Networks and LSTMs for Segmentation Free OCR},
  booktitle = {Proc. ICDAR},
  year      = {2017},
  url       = {https://doi.org/10.1109/ICDAR.2017.34},
  doi       = {10.1109/ICDAR.2017.34}
}