A thermal Object Tracking benchmark

Abstract

Short-term single-object (STSO) tracking in thermal images is a challenging problem relevant in a growing number of applications. In order to evaluate STSO tracking algorithms on visual imagery, there are de facto standard benchmarks. However, we argue that tracking in thermal imagery is different than in visual imagery, and that a separate benchmark is needed. The available thermal infrared datasets are few and the existing ones are not challenging for modern tracking algorithms. Therefore, we hereby propose a thermal infrared benchmark according to the Visual Visual Object Tracking (VOT) protocol for evaluation of STSO tracking methods. The benchmark includes the new LTIR dataset containing 20 thermal image sequences which have been collected from multiple sources and annotated in the format used in the VOT Challenge. In addition, we show that the ranking of different tracking principles differ between the visual and thermal benchmarks, confirming the need for the new benchmark.

Evaluating template rescaling in short-term single-object tracking

“…Indicators of the popularity of the topic are challenges/contest like the recurring Visual Object Tracking (VOT) challenge [5,6], the Linköping Thermal IR tracking benchmark (LTIR) [1], the Online Object Tracking (OTB) benchmark [8], and the series of workshops on Performance Evaluation of Tracking and Surveillance (PETS) [9].…”

The Thermal Infrared Visual Object Tracking VOT-TIR2016 Challenge Results

“…Available datasets for evaluation of tracking in thermal infrared have become outdated [3]. This causes researchers to evaluate their methods on proprietary datasets, which makes it difficult to get an overview of advancement made in the field.…”

“…Further, due to the limited amount of freely available thermal infrared datasets and sequences, sequence selection could not be done as in VOT2015. A new dataset, LTIR (the Linköping Thermal IR dataset) [3], was created for this purpose. Seven different sources were asked to contribute with data and the provided data that contained sufficiently challenging tracking events were included in the dataset.…”

“…The dataset used in VOT-TIR2015 is LTIR, the Linköping Thermal IR dataset [3]. Sequences included in the dataset were collected from seven different sources using eight different types of sensors.…”

“…The Visual Object Tracking challenge, provides an established evaluation methodology for data in the visible spectrum. The main idea of VOT-TIR2015 is to carry these ideas to the area of TIR data, based on a recently collected dataset [3].…”