TY - JOUR
T1 - A Novel Performance Evaluation Methodology for Single-Target Trackers
AU - Kristan, Matej
AU - Matas, Jiri
AU - Leonardis, Ales
AU - Vojir, Tomas
AU - Pflugfelder, Roman
AU - Fernandez, Gustavo
AU - Nebehay, Georg
AU - Porikli, Fatih
AU - Cehovin, Luka
N1 - Publisher Copyright:
© 2016 IEEE.
PY - 2016/11/1
Y1 - 2016/11/1
N2 - This paper addresses the problem of single-target tracker performance evaluation. We consider the performance measures, the dataset and the evaluation system to be the most important components of tracker evaluation and propose requirements for each of them. The requirements are the basis of a new evaluation methodology that aims at a simple and easily interpretable tracker comparison. The ranking-based methodology addresses tracker equivalence in terms of statistical significance and practical differences. A fully-annotated dataset with per-frame annotations with several visual attributes is introduced. The diversity of its visual properties is maximized in a novel way by clustering a large number of videos according to their visual attributes. This makes it the most sophistically constructed and annotated dataset to date. A multi-platform evaluation system allowing easy integration of third-party trackers is presented as well. The proposed evaluation methodology was tested on the VOT2014 challenge on the new dataset and 38 trackers, making it the largest benchmark to date. Most of the tested trackers are indeed state-of-the-art since they outperform the standard baselines, resulting in a highly-challenging benchmark. An exhaustive analysis of the dataset from the perspective of tracking difficulty is carried out. To facilitate tracker comparison a new performance visualization technique is proposed.
AB - This paper addresses the problem of single-target tracker performance evaluation. We consider the performance measures, the dataset and the evaluation system to be the most important components of tracker evaluation and propose requirements for each of them. The requirements are the basis of a new evaluation methodology that aims at a simple and easily interpretable tracker comparison. The ranking-based methodology addresses tracker equivalence in terms of statistical significance and practical differences. A fully-annotated dataset with per-frame annotations with several visual attributes is introduced. The diversity of its visual properties is maximized in a novel way by clustering a large number of videos according to their visual attributes. This makes it the most sophistically constructed and annotated dataset to date. A multi-platform evaluation system allowing easy integration of third-party trackers is presented as well. The proposed evaluation methodology was tested on the VOT2014 challenge on the new dataset and 38 trackers, making it the largest benchmark to date. Most of the tested trackers are indeed state-of-the-art since they outperform the standard baselines, resulting in a highly-challenging benchmark. An exhaustive analysis of the dataset from the perspective of tracking difficulty is carried out. To facilitate tracker comparison a new performance visualization technique is proposed.
KW - Performance analysis
KW - model-free tracking
KW - single-target tracking
KW - tracker evaluation datasets
KW - tracker evaluation methodology
KW - tracker evaluation system
UR - http://www.scopus.com/inward/record.url?scp=84992034892&partnerID=8YFLogxK
U2 - 10.1109/TPAMI.2016.2516982
DO - 10.1109/TPAMI.2016.2516982
M3 - Article
SN - 0162-8828
VL - 38
SP - 2137
EP - 2155
JO - IEEE Transactions on Pattern Analysis and Machine Intelligence
JF - IEEE Transactions on Pattern Analysis and Machine Intelligence
IS - 11
M1 - 7379002
ER -