Welcome to UNAGI’s documentation!

UNAGI (U-Net Adaptive Generalized Image Binarization) is a python package to perform binarization on document images to improve OCR results. UNAGI performs segmantic segmentation on the image and creates a binary image. U-net is used as the model architecture to train the deep learning model. UNAGI comes with a pretrained model trained on DIBCO dataset.

Tutorials

• Quick Start Guide
  • Installation
  • Features
  • Example Usage
  • Tips
  • Notes
  • Warnings
• unagi
  • Modules
  • Sub Packages

In-depth

• Artificial Neural Networks
• Optimization of Nerual Networks
• Convolutional Neural Networks
• Semantic Segmentation
• Optical Character Recognition

Indices and tables

• Index

• Module Index

• Search Page