Image Ground Truth GUI Tools

Posted on St 11 září 2013 in misc

Evaluation of computer vision algorithms usually requires comparison against manually created ground truth. Following two GUI tools are easy to use for a basic annotation in images. Both are written in Python and should be easy to modify.

pilab-annotator

pilab-annotator screenshot

http://code.google.com/p/pilab-annotator/

features:

  • works with image series
  • saves annotated objects to xml files
  • supports annotating of point and rectangle objects

pylabelme

pylabelme screenshot

https://github.com/mpitid/pylabelme

features:

  • works with a single image
  • supports polygon annotations
  • saves annotated objects to json files

Pylableme workflow is demonstrated in the following screencast.

python