(WIP) Julia and Flux adaptation of the python package aocr.py https://github.com/emedvedev/attention-ocr using Seq2seq with Attention from https://github.com/merckxiaan/flux-seq2seq
The scope at the moment is only numeric digits 0-9
- Package pre-compiles successfully
- Flux model is able to train but slowly
- Add predict function
- Add unemmebeding function to take predictions from vectors to digits
- Test that the model works on example data used successfully with aocr.py