Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Thermal MOT: A Novel Box Association Method Leveraging Thermal Identity and Motion Similarity

Wassim El Ahmar, Dhanvin Kolhatkar, Farzan Nowruzi, Robert Laganiere

TL;DR

This paper introduces an innovative approach to improve MOT in the thermal domain by developing a novel box association method that utilizes both thermal object identity and motion similarity, enabling more accurate and robust MOT performance.

Abstract

Multiple Object Tracking (MOT) in thermal imaging presents unique challenges due to the lack of visual features and the complexity of motion patterns. This paper introduces an innovative approach to improve MOT in the thermal domain by developing a novel box association method that utilizes both thermal object identity and motion similarity. Our method merges thermal feature sparsity and dynamic object tracking, enabling more accurate and robust MOT performance. Additionally, we present a new dataset comprised of a large-scale collection of thermal and RGB images captured in diverse urban environments, serving as both a benchmark for our method and a new resource for thermal imaging. We conduct extensive experiments to demonstrate the superiority of our approach over existing methods, showing significant improvements in tracking accuracy and robustness under various conditions. Our findings suggest that incorporating thermal identity with motion data enhances MOT performance. The newly collected dataset and source code is available at https://github.com/wassimea/thermalMOT

Enhancing Thermal MOT: A Novel Box Association Method Leveraging Thermal Identity and Motion Similarity

TL;DR

This paper introduces an innovative approach to improve MOT in the thermal domain by developing a novel box association method that utilizes both thermal object identity and motion similarity, enabling more accurate and robust MOT performance.

Abstract

Multiple Object Tracking (MOT) in thermal imaging presents unique challenges due to the lack of visual features and the complexity of motion patterns. This paper introduces an innovative approach to improve MOT in the thermal domain by developing a novel box association method that utilizes both thermal object identity and motion similarity. Our method merges thermal feature sparsity and dynamic object tracking, enabling more accurate and robust MOT performance. Additionally, we present a new dataset comprised of a large-scale collection of thermal and RGB images captured in diverse urban environments, serving as both a benchmark for our method and a new resource for thermal imaging. We conduct extensive experiments to demonstrate the superiority of our approach over existing methods, showing significant improvements in tracking accuracy and robustness under various conditions. Our findings suggest that incorporating thermal identity with motion data enhances MOT performance. The newly collected dataset and source code is available at https://github.com/wassimea/thermalMOT

Paper Structure

This paper contains 23 sections, 3 equations, 3 figures, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Multiple Object Trackers
  3. Contributions
  4. Literature Review
  5. Object Detection
  6. Two-Stage Multiple Object Trackers
  7. Thermal Sensors
  8. Datasets
  9. Data Collection
  10. Experiments
  11. Benchmarking SOTA MOT models on the RGB-Thermal MOT dataset
  12. Development of New MOT Box Association Method Utilizing Thermal Information
  13. Thermal and Motion Bounding Boxes:
  14. Initialization:
  15. Similarity Matrix Construction:
  16. ...and 8 more sections

Figures (3)

  • Figure 1: Sample RGB and corresponding thermal images from the RGB-Thermal MOT dataset.
  • Figure 2: Left: MOTA and IDF1 values of our proposed box association method used with ByteTrack. Right: MOTA and IDF1 values used with OCSORT on the validation sequences of the RGB-Thermal MOT dataset using different values of alpha.
  • Figure 3: Overall comparison of the important MOT metrics running standard motion box association against our proposed box association algorithm using ByteTrack.