Multiple object ground truth annotation tool using polygons
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
skcf
vivalib
.gitignore
CMakeLists.txt
README.md
actionType.txt.example
aprocess.cpp
aprocess.h
clp.cpp
clp.hpp
extract.cpp
license
main.cpp
mprocessor.cpp
mprocessor.hpp
output.xml.example.xml
rapidxml.hpp
rapidxml_print.hpp

README.md

Annotate is a multiple object ground truth annotation tool using polygons.

The Annotate tool allows you to create your video sequence annotations using polygonal representations. It generates a text file with the annotations for each frame following a simple CSV format.

Among the possible applications Annotate could be useful for creating ground truth for:

  1. Pedestrian, people or re-identification datasets
  2. Upper body, face or other parts datasets
  3. Object detection datasets
  4. Object tracking datasets

In general, any project where a polygonal representation of your target is needed, Annotate will help to create it.

The Annotate tool can also provide a prediction for target positions to help annotating large datasets.

For more detailed information and examples of how to use reefer to the project's wiki.

Contributors

Thanks to:

  1. Dongfeng Gu - Adding write/read support for simplified XML version format.