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:
- Pedestrian, people or re-identification datasets
- Upper body, face or other parts datasets
- Object detection datasets
- 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.
Thanks to:
- Dongfeng Gu - Adding write/read support for simplified XML version format.