In order to train a model, data must be downloaded and placed in this directory.
Follow the download instructions here http://www.fki.inf.unibe.ch/databases/iam-on-line-handwriting-database.
Files required are:
- ascii-all.tar.gz - https://fki.tic.heia-fr.ch/DBs/iamOnDB/data/ascii-all.tar.gz
- lineStrokes-all.tar.gz - https://fki.tic.heia-fr.ch/DBs/iamOnDB/data/lineStrokes-all.tar.gz
- original-xml-part.tar.gz - https://fki.tic.heia-fr.ch/DBs/iamOnDB/data/original-xml-part.tar.gz
Only a subset of the downloaded data is required. Move the relevant download data so the directory structure is as folllows:
data/
├── raw/
│ ├── ascii/
│ ├── lineStrokes/
│ ├── original/
| blacklist.npy
Once this is completed, run
from handwriting_synthesis.training.preparation import prepare
prepare()extract the data and dump it to numpy files.
To train the model, run
from handwriting_synthesis.training import train
train()This takes a couple of days on a single Tesla K80.