The toolkit is used to evaluate visual trackers - it implements VOT measures accuracy and robustness on a baseline (reset-based) experiment. The toolkit is tested on Matlab and Octave and it is compatible with VOT sequences.
Note: The toolkit is developed for educational purposes only (course Advanced Computer Vision Methods at Faculty of Computer and information Science, University of Ljubljana) and it should not be used in research community.
1.) Checkout the repository in an empty folder - called toolkit
2.) Create new folder (not within toolkit
) - called workspace
3.) Add the toolkit
path to Matlab paths
4.) Run command (within workspace
folder): create_tracker_workspace
5.) A script called workspace_config
will appear in the workspace
. Set the following variables according to your settings in this script:
c.tracker_name = ‘<tracker-name>’;
c.additional_paths = {‘add-paths’}; % Path to the source code directory
c.dataset_path = ‘path-to-dataset-directory’;
Implement the tracker (with name <tracker-name>
) in two functions:
tracker = <tracker-name>_initialize(img, bbox)
[tracker, bbox] = <tracker-name>_update(tracker, img)
You can see an example for the NCC tracker (located in examples/ncc
folder).
In the workspace
folder run command: run_experiment
. The toolkit will run the tracker on all sequences and save results in the results folder. After the experiment is over you can call:
performance_summary
Tracking performance of the tracker specified inworkspace_config
will be printed out and a latex table will be generated in output folder.compare_trackers({‘list-comma-separated-tracker-names’})
Generates similar latex table asperformance_summary
script, but with multiple columns corresponding to multiple trackers. Additionally, an A-R plot will be generated and stored in output folder.