Changelog

History

0.4.1 (2022-09-01)

Changes:

• Drop coco-ash dependency. coco-ash was a personal fork of pycocotools and was created as a workaround for when poetry did not support git dependencies rooted in subdirectories. Since that is no longer the case, we're dropping the coco-ash dependency.

0.4.0 (2021-08-23)

Changes:

• Drop Python 3.6 support
• Improved documentation
• Standardise docstring styles (#32)

Bugfixes:

• Bugfix for merge creating duplicated IDs by philipsgithub (#50)

0.3.5 (2021-05-06)

Changes:

• Replace os with pathlib for managing paths.
• Fixed bug in det2seg where small annotations were covered by larger annotations.
• Faster category removal by @xychen9459 (#43).

0.3.4 (2021-03-14)

Changes:

• Add support for merging datasets with different categories.
• Adding colour palette to segmasks is now optional.
• Ignore hidden files.

0.3.1 (2020-05-18)

Bugfix:

• Fix for coco_stats by @Lplenka. Refer #15

0.3.0 (2020-04-19)

Changes:

• combine has been renamed to merge.
• Add support for merging annotations only.

Deprecations:

• Deprecate converters. This was motivated by two reasons:
  • There are too many conversion formats. Trying to include even the most popular ones makes the project unwieldy. Better to use one of the readily available scripts online.
  • Until now, the repository only supported TFRecord. Tensorflow is a large library and it does not make sense for it to be a project requirement when it's only used for a singular task of converting to TFRecord.

0.2.0 (2019-11-28)

New features:

• Generate anchors from the dataset using K-means.
• Generate segmentation masks from dataset.

Changes:

• Modified category removal to accept a list of categories as opposed to entering categories one after the other.

0.1.0 (2019-10-07)

First release on PyPI. Supports the following functions:

• Merge datasets.
• Remove specfiic category from dataset.
• Generate annotations statistics - distribution of object areas and category distribution.
• Annotation visualiser for viewing the entire dataset.
• Converter for converting annotations to TFRecord format.