start_follow_read_py3

This repository is the implementation of the methods described in the paper Start, Follow, Read: Full-Page End-to-end Handwriting Recognition. The code is heavily based on the original code as released by the authors. Our repository has some minor changes to make it run with python3 and pytorch=1.3.

Dependencies

The dependencies are all found in sfr_env_environment.yaml. They are installed as follows.

conda env create -f sfr_env_environment.yaml

The environment is activated as source activate sfr_env.

Running the code

All steps to run the code are same as in official code released at https://github.com/cwig/start_follow_read