Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Automatic Data Reduction of Image Sequences Acquired in Object Tracking Mode for Detection and Position Measurement of Faint Orbital Objects.

Sensors (Basel, Switzerland)·2026
Same author

Forest Inspection Dataset: A Synthetic UAV Dataset for Semantic Segmentation of Forest Environments.

Scientific data·2026
Same author

Fully Convolutional Neural Network for Vehicle Speed and Emergency-Brake Prediction.

Sensors (Basel, Switzerland)·2024
Same author

Hepatocellular Carcinoma Recognition from Ultrasound Images Using Combinations of Conventional and Deep Learning Techniques.

Sensors (Basel, Switzerland)·2023
Same author

Enhanced Perception for Autonomous Driving Using Semantic and Geometric Data Fusion.

Sensors (Basel, Switzerland)·2022
Same author

Part-Based Obstacle Detection Using a Multiple Output Neural Network.

Sensors (Basel, Switzerland)·2022

Related Experiment Video

Updated: Oct 10, 2025

Trajectory Data Analyses for Pedestrian Space-time Activity Study
16:14

Trajectory Data Analyses for Pedestrian Space-time Activity Study

Published on: February 25, 2013

13.7K

Robust Data Association Using Fusion of Data-Driven and Engineered Features for Real-Time Pedestrian Tracking in

Mircea Paul Muresan1, Sergiu Nedevschi1, Radu Danescu1

  • 1Computer Science Department, Technical University of Cluj-Napoca, 28 Memorandumului Street, 400114 Cluj Napoca, Romania.

Sensors (Basel, Switzerland)
|December 10, 2021
PubMed
Summary

This study enhances multi-object tracking in thermal images using Siamese Networks and an edge-based descriptor. The combined approach improves data association for robust and adaptable object tracking solutions.

Keywords:
advanced driving assistance systemsautonomous drivingconvolutional neural networksdata association and trackingfeature engineeringthermal imaging

More Related Videos

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
08:25

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Published on: May 7, 2019

9.2K
FIM Imaging and FIMtrack: Two New Tools Allowing High-throughput and Cost Effective Locomotion Analysis
10:02

FIM Imaging and FIMtrack: Two New Tools Allowing High-throughput and Cost Effective Locomotion Analysis

Published on: December 24, 2014

11.9K

Related Experiment Videos

Last Updated: Oct 10, 2025

Trajectory Data Analyses for Pedestrian Space-time Activity Study
16:14

Trajectory Data Analyses for Pedestrian Space-time Activity Study

Published on: February 25, 2013

13.7K
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
08:25

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Published on: May 7, 2019

9.2K
FIM Imaging and FIMtrack: Two New Tools Allowing High-throughput and Cost Effective Locomotion Analysis
10:02

FIM Imaging and FIMtrack: Two New Tools Allowing High-throughput and Cost Effective Locomotion Analysis

Published on: December 24, 2014

11.9K

Area of Science:

  • Computer Vision
  • Machine Learning

Background:

  • Object tracking in thermal images presents challenges due to low resolution, lack of color, and high inter-class similarity.
  • The data association problem, crucial for multi-object tracking, involves matching detections to existing tracks and handling appearance changes.

Purpose of the Study:

  • To improve data association for multi-object tracking in thermal imagery.
  • To develop a robust and adaptable tracking solution by combining data-driven and feature-engineered approaches.

Main Methods:

  • Developed a data-driven appearance score using five Siamese Networks operating on image detections.
  • Engineered an original edge-based descriptor to enhance the data association process.
  • Created a new dataset of pedestrian instances for training the Siamese Networks.

Main Results:

  • Achieved an average precision of 86.2% on publicly available benchmarks.
  • Demonstrated a running time of 25 ms, indicating efficient processing.
  • The combination of data-driven scores and feature engineering resulted in a powerful tracking solution.

Conclusions:

  • The proposed method effectively addresses the data association challenge in thermal object tracking.
  • The integration of Siamese Networks and edge-based descriptors provides robustness and adaptability for diverse scenarios.
  • The developed dataset aids in training and evaluating thermal object tracking algorithms.