Home > Published Issues > 2022 > Volume 10, No. 2, June 2022 >
JOIG 2022 Vol.10(2): 88-94
doi: 10.18178/joig.10.2.88-94

Performance Analysis of Fuzzy-Weighted Multiple Instance Learning on Thermal Video-Based Visual Tracking

Nur Ibrahim, Arsyad R. Darlis, and Benyamin Kusumoputro
Department of Electrical Engineering Universitas Indonesia, Depok 16424, Indonesia

Abstract—In this paper, performance analysis of fuzzy-Weighted Multiple Instance Learning (WMIL) with the fuzzy logic tracker on thermal video-based visual tracking is presented. Thermal cameras have been used recently in some pedestrian areas, cars, and surveillance areas that need to be monitored all day. A thermal camera with advantages over the other visual-based sensors in low-light conditions is utilized in this research. The paper presents an analysis of visual tracking with an experimental method in the low-light outdoor environment. The thermal camera is used to record object movement used as video sequences to analyze the performance of our proposed system that integrate Type-2 Fuzzy Logic System (T2FLS) and WMIL tracker. The WMIL-T2FLS tracker performance is shown in the failure rate and center location error. The results show that the object in the thermal video sequences can be tracked using WMIL-T2FLS tracker in the low-light outdoor environment with a low level of failure rate and center location error. Then, the WMIL-T2FLS tracker can track the object when it occluded with the other similar object quite accurately. This result was compared with the original WMIL and some state-of-the art of tracking algorithm: DSST, ECO, KCF, SRDCF, and BACF. The research results showed that the WMIL-T2FLS system significantly improved compared with the WMIL method only, with a success rate improvement of at least 35 % and precision of at least 0.2 in 15 m dan 10 m. WMIL-T2FLS tracker also outperform some state-of-the art method and showed good performance in visual tracking at low-light environments.

Index Terms—thermal camera, type-2 fuzzy logic system, visual tracking, weighted multiple instance learning, low-light outdoor environment

Cite: Nur Ibrahim, Arsyad R. Darlis, and Benyamin Kusumoputro, "Performance Analysis of Fuzzy-Weighted Multiple Instance Learning on Thermal Video-Based Visual Tracking," Journal of Image and Graphics, Vol. 10, No. 2, pp. 88-94, June 2022.

Copyright © 2022 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.