Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Learning Traffic Anomalies from Generative Models on Real-Time Observations

Giasemis, Fotis I. and Sopasakis, Alexandros LU orcid (2025) 10th International Conference on Signal and Image Processing, ICSIP 2025 In 2025 10th International Conference on Signal and Image Processing, ICSIP 2025 p.427-432
Abstract

Accurate detection of traffic anomalies is crucial for effective urban traffic management and congestion mitigation. We use the Spatiotemporal Generative Adversarial Network (STGAN) framework combining Graph Neural Networks and Long Short-Term Memory networks to capture complex spatial and temporal dependencies in traffic data. We apply STGAN to real-time, minute-by-minute observations from 42 traffic cameras across Gothenburg, Sweden, collected over several months in 2020. The images are processed to compute a flow metric representing vehicle density, which serves as input for the model. Training is conducted on data from April to November 2020, and validation is performed on a separate dataset from November 14 to 23, 2020. Our results... (More)

Accurate detection of traffic anomalies is crucial for effective urban traffic management and congestion mitigation. We use the Spatiotemporal Generative Adversarial Network (STGAN) framework combining Graph Neural Networks and Long Short-Term Memory networks to capture complex spatial and temporal dependencies in traffic data. We apply STGAN to real-time, minute-by-minute observations from 42 traffic cameras across Gothenburg, Sweden, collected over several months in 2020. The images are processed to compute a flow metric representing vehicle density, which serves as input for the model. Training is conducted on data from April to November 2020, and validation is performed on a separate dataset from November 14 to 23, 2020. Our results demonstrate that the model effectively detects traffic anomalies with high precision and low false positive rates. The detected anomalies include camera signal interruptions, visual artifacts, and extreme weather conditions affecting traffic flow.

(Less)
Please use this url to cite or link to this publication:
author
and
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
Generative Adversarial Networks, Graph Neural Networks, Spatiotemporal Modeling, Traffic Anomaly Detection, Urban Traffic Management
host publication
2025 10th International Conference on Signal and Image Processing, ICSIP 2025
series title
2025 10th International Conference on Signal and Image Processing, ICSIP 2025
pages
6 pages
publisher
IEEE - Institute of Electrical and Electronics Engineers Inc.
conference name
10th International Conference on Signal and Image Processing, ICSIP 2025
conference location
Wuxi, China
conference dates
2025-07-12 - 2025-07-14
external identifiers
  • scopus:105019520426
ISBN
9798331536992
DOI
10.1109/ICSIP65915.2025.11171552
language
English
LU publication?
yes
additional info
Publisher Copyright: © 2025 IEEE.
id
9952a8eb-a0dc-4bc8-a55a-faec277bf100
date added to LUP
2025-10-30 05:08:38
date last changed
2026-02-06 15:02:19
@inproceedings{9952a8eb-a0dc-4bc8-a55a-faec277bf100,
  abstract     = {{<p>Accurate detection of traffic anomalies is crucial for effective urban traffic management and congestion mitigation. We use the Spatiotemporal Generative Adversarial Network (STGAN) framework combining Graph Neural Networks and Long Short-Term Memory networks to capture complex spatial and temporal dependencies in traffic data. We apply STGAN to real-time, minute-by-minute observations from 42 traffic cameras across Gothenburg, Sweden, collected over several months in 2020. The images are processed to compute a flow metric representing vehicle density, which serves as input for the model. Training is conducted on data from April to November 2020, and validation is performed on a separate dataset from November 14 to 23, 2020. Our results demonstrate that the model effectively detects traffic anomalies with high precision and low false positive rates. The detected anomalies include camera signal interruptions, visual artifacts, and extreme weather conditions affecting traffic flow.</p>}},
  author       = {{Giasemis, Fotis I. and Sopasakis, Alexandros}},
  booktitle    = {{2025 10th International Conference on Signal and Image Processing, ICSIP 2025}},
  isbn         = {{9798331536992}},
  keywords     = {{Generative Adversarial Networks; Graph Neural Networks; Spatiotemporal Modeling; Traffic Anomaly Detection; Urban Traffic Management}},
  language     = {{eng}},
  pages        = {{427--432}},
  publisher    = {{IEEE - Institute of Electrical and Electronics Engineers Inc.}},
  series       = {{2025 10th International Conference on Signal and Image Processing, ICSIP 2025}},
  title        = {{Learning Traffic Anomalies from Generative Models on Real-Time Observations}},
  url          = {{http://dx.doi.org/10.1109/ICSIP65915.2025.11171552}},
  doi          = {{10.1109/ICSIP65915.2025.11171552}},
  year         = {{2025}},
}