Overcoming Data Scarcity in Roadside Thermal Imagery: A New Dataset and Weakly Supervised Incremental Learning Framework

Roadside camera systems are commonly used for traffic data collection, yet conventional optical systems are limited by poor performance in varying weather and light conditions and are often restricted by data privacy regulations. Thermal imaging overcomes these issues, enabling reliable detection ac...

Full description

Saved in:
Bibliographic Details
Main Authors: Arnd Pettirsch, Alvaro Garcia-Hernandez
Format: Article
Language:English
Published: MDPI AG 2025-04-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/7/2340
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849730264246255616
author Arnd Pettirsch
Alvaro Garcia-Hernandez
author_facet Arnd Pettirsch
Alvaro Garcia-Hernandez
author_sort Arnd Pettirsch
collection DOAJ
description Roadside camera systems are commonly used for traffic data collection, yet conventional optical systems are limited by poor performance in varying weather and light conditions and are often restricted by data privacy regulations. Thermal imaging overcomes these issues, enabling reliable detection across all conditions without collecting personal data. However, its widespread use is hindered by the scarcity of diverse, annotated thermal training data, especially since fixed cameras installed at the side of the road produce very similar images with the same backgrounds. This paper presents two key innovations to address these challenges: a novel dataset of 11,400 annotated images and 142 unannotated video clips, the largest and most diverse available for thermal roadside imaging to date, and a weakly supervised incremental learning framework tailored for thermal roadside imagery. The dataset supports the development of self-supervised algorithms, and the learning framework allows efficient adaptation to new camera viewpoints and diverse environmental conditions without additional labelling. Together, these contributions enable cost-effective and reliable thermal-based traffic monitoring across varied locations, achieving an 8.9-point increase in mean average precision for previously unseen viewpoints.
format Article
id doaj-art-7d34dcff476344c9866f979f2f5e3f27
institution DOAJ
issn 1424-8220
language English
publishDate 2025-04-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-7d34dcff476344c9866f979f2f5e3f272025-08-20T03:08:56ZengMDPI AGSensors1424-82202025-04-01257234010.3390/s25072340Overcoming Data Scarcity in Roadside Thermal Imagery: A New Dataset and Weakly Supervised Incremental Learning FrameworkArnd Pettirsch0Alvaro Garcia-Hernandez1Institute for Highway Engineering, RWTH Aachen University, 52062 Aachen, GermanyInstitute for Highway Engineering, RWTH Aachen University, 52062 Aachen, GermanyRoadside camera systems are commonly used for traffic data collection, yet conventional optical systems are limited by poor performance in varying weather and light conditions and are often restricted by data privacy regulations. Thermal imaging overcomes these issues, enabling reliable detection across all conditions without collecting personal data. However, its widespread use is hindered by the scarcity of diverse, annotated thermal training data, especially since fixed cameras installed at the side of the road produce very similar images with the same backgrounds. This paper presents two key innovations to address these challenges: a novel dataset of 11,400 annotated images and 142 unannotated video clips, the largest and most diverse available for thermal roadside imaging to date, and a weakly supervised incremental learning framework tailored for thermal roadside imagery. The dataset supports the development of self-supervised algorithms, and the learning framework allows efficient adaptation to new camera viewpoints and diverse environmental conditions without additional labelling. Together, these contributions enable cost-effective and reliable thermal-based traffic monitoring across varied locations, achieving an 8.9-point increase in mean average precision for previously unseen viewpoints.https://www.mdpi.com/1424-8220/25/7/2340thermal imagerytraffic data collectionroadside camerasweakly supervised learningincremental learningthermal image dataset
spellingShingle Arnd Pettirsch
Alvaro Garcia-Hernandez
Overcoming Data Scarcity in Roadside Thermal Imagery: A New Dataset and Weakly Supervised Incremental Learning Framework
Sensors
thermal imagery
traffic data collection
roadside cameras
weakly supervised learning
incremental learning
thermal image dataset
title Overcoming Data Scarcity in Roadside Thermal Imagery: A New Dataset and Weakly Supervised Incremental Learning Framework
title_full Overcoming Data Scarcity in Roadside Thermal Imagery: A New Dataset and Weakly Supervised Incremental Learning Framework
title_fullStr Overcoming Data Scarcity in Roadside Thermal Imagery: A New Dataset and Weakly Supervised Incremental Learning Framework
title_full_unstemmed Overcoming Data Scarcity in Roadside Thermal Imagery: A New Dataset and Weakly Supervised Incremental Learning Framework
title_short Overcoming Data Scarcity in Roadside Thermal Imagery: A New Dataset and Weakly Supervised Incremental Learning Framework
title_sort overcoming data scarcity in roadside thermal imagery a new dataset and weakly supervised incremental learning framework
topic thermal imagery
traffic data collection
roadside cameras
weakly supervised learning
incremental learning
thermal image dataset
url https://www.mdpi.com/1424-8220/25/7/2340
work_keys_str_mv AT arndpettirsch overcomingdatascarcityinroadsidethermalimageryanewdatasetandweaklysupervisedincrementallearningframework
AT alvarogarciahernandez overcomingdatascarcityinroadsidethermalimageryanewdatasetandweaklysupervisedincrementallearningframework