The analysis module contains functions related to performance analysis and general analysis of results. The core of the module is a ranking methodology that is based on Accuracy-Robustness measure pair, proposed in this paper.
- estimate_accuracy - Calculate accuracy score
- estimate_failures - Computes number of failures score
- estimate_expected_overlap - Estimates expected average overlap for different sequence lengths
- compare_trackers - Compares two trackers in terms of accuracy and robustness
- adapted_ranks - Performs rank adaptation on a set of ranks
- analyze_ranks - Performs ranking analysis on per-sequence or per-tag basis
- create_tag_selectors - Create per-tag selectors
- create_sequence_selectors - Create per-sequence selectors
- normalize_speed - Normalizes tracker speed estimate
- analyze_speed - Perform speed analysis
- analyze_expected_overlap - Performs expected overlap analysis
- analyze_failures - Perform failure frequency analysis