End-to-End Handwritten Text Detection and Transcription in Full Pages

Pytorch implementation of the paper End-To-End Handwritten Text Detection and Transcription in Full Pages by Manuel Carbonell, Joan Mas, Mauricio Villegas, Alicia Fornés and Josep Lladós.

Installation

1. Clone this repo

2. Install the required packages:

apt-get install tk-dev python-tk

3. Install the python packages:

pip3 install cffi

pip3 install cython

pip3 install opencv-python

pip3 install requests

pip3 install editdistance

Warp ctc pytorch: https://github.com/SeanNaren/warp-ctc

Python pagexml: https://github.com/omni-us/pagexml/tree/master/py-pagexml

Data

Train, validation and test examples's ground truth boxes and text contents must be listed in a csv file with the following format:

path/to/image1.png,text_bbox_1_x0,text_bbox_1_y0,text_bbox_1_x1,text_bbox_1_y1,'text',text_bbox_1_content
path/to/image1.png,text_bbox_2_x0,text_bbox_2_y0,text_bbox_2_x1,text_bbox_2_y1,'text',text_bbox_2_content
...
path/to/imageN.png,text_bbox_1_x0,text_bbox_1_y0,text_bbox_1_x1,text_bbox_1_y1,'text',text_bbox_1_content

Training

./experiment_e2e.sh