Image-dataset-converter-paddle

------------------

- added `--resume_from` option to relevant readers that allows resuming the data processing
from the first file that matches this glob expression (e.g., `*/012345.txt`)
- the `from-paddle-is` reader and `to-paddle-is` writer support custom separators for the text
files linking image and annotation, e.g., one can use a semi-colon instead of the default
space when file names should contain spaces as well
(use `separator: ;` in the `train_dataset` and `val_dataset` sections)
- the `to-paddle-ic` writer now raises an exception if it encounters a space in an image name
- new grouping support added to writers via the `--split_group` option

------------------

- added placeholder support: `from-paddle-ic`, `from-paddle-is`, `to-paddle-ic`, and `to-paddle-is`

------------------

- initial release



PyPi
====

Preparation:

* update all help files (`img-help -f markdown -o plugins -t "paddle plugins" -i README.md -l INFO`)
* increment version in `setup.py`
* add new changelog section in `CHANGES.rst`
* align `DESCRIPTION.rst` with `README.md`
* commit/push all changes

Commands for releasing on pypi.org (requires twine >= 1.8.0):


find -name "*~" -delete
rm dist/*
python3 setup.py clean
python3 setup.py sdist
twine upload dist/*



Github
======

Steps:

* start new release (version: `vX.Y.Z`)
* enter release notes, i.e., significant changes since last release
* upload `image_dataset_converter_paddle-X.Y.Z.tar.gz` previously generated with `setup.py`
* publish


image-dataset-converter-all
===========================

* increment minimum version to newly released one in `setup.py`
* add note to `CHANGES.rst`