Integrating Viewing Direction and Image Features for Robust Multi-View Multi-Object 3D Pedestrian Tracking

Recently, there has been growing interest in the development of 3D multi-view, multi-object detection and tracking models (MV-MOD and MV-MOT), resulting in significant methodological advances. However, many of these developments do not address the critical challenge of generalization across differen...

Full description

Saved in:
Bibliographic Details
Main Authors: R. Ali, M. Mehltretter, C. Heipke
Format: Article
Language:English
Published: Copernicus Publications 2025-07-01
Series:ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Online Access:https://isprs-annals.copernicus.org/articles/X-G-2025/47/2025/isprs-annals-X-G-2025-47-2025.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849321207519772672
author R. Ali
M. Mehltretter
C. Heipke
author_facet R. Ali
M. Mehltretter
C. Heipke
author_sort R. Ali
collection DOAJ
description Recently, there has been growing interest in the development of 3D multi-view, multi-object detection and tracking models (MV-MOD and MV-MOT), resulting in significant methodological advances. However, many of these developments do not address the critical challenge of generalization across different camera constellations, i.e., having camera constellations that differ between training and testing, limiting their effectiveness in real-world applications. A key factor often overlooked is the influence of the direction of the optical axis during image capture, which is not adequately propagated in the model. In this work, we propose a novel convolutional neural network-based method for 3D MV-MOD and MV-MOT that enhances generalization by incorporating the direction from which the images were captured as an additional input to this network. For each image, this directional information is combined with the 2D features extracted from that image, before 3D features are computed, using the 2D features from all images. We empirically evaluate the performance of the proposed method on the real-world Wildtrack dataset, demonstrating the effectiveness of the proposed approach.
format Article
id doaj-art-ef7e6f57d52a402c85f07a8403f65638
institution Kabale University
issn 2194-9042
2194-9050
language English
publishDate 2025-07-01
publisher Copernicus Publications
record_format Article
series ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
spelling doaj-art-ef7e6f57d52a402c85f07a8403f656382025-08-20T03:49:49ZengCopernicus PublicationsISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences2194-90422194-90502025-07-01X-G-2025475510.5194/isprs-annals-X-G-2025-47-2025Integrating Viewing Direction and Image Features for Robust Multi-View Multi-Object 3D Pedestrian TrackingR. Ali0M. Mehltretter1C. Heipke2Institute of Photogrammetry and GeoInformation, Leibniz University Hannover, GermanyInstitute of Photogrammetry and GeoInformation, Leibniz University Hannover, GermanyInstitute of Photogrammetry and GeoInformation, Leibniz University Hannover, GermanyRecently, there has been growing interest in the development of 3D multi-view, multi-object detection and tracking models (MV-MOD and MV-MOT), resulting in significant methodological advances. However, many of these developments do not address the critical challenge of generalization across different camera constellations, i.e., having camera constellations that differ between training and testing, limiting their effectiveness in real-world applications. A key factor often overlooked is the influence of the direction of the optical axis during image capture, which is not adequately propagated in the model. In this work, we propose a novel convolutional neural network-based method for 3D MV-MOD and MV-MOT that enhances generalization by incorporating the direction from which the images were captured as an additional input to this network. For each image, this directional information is combined with the 2D features extracted from that image, before 3D features are computed, using the 2D features from all images. We empirically evaluate the performance of the proposed method on the real-world Wildtrack dataset, demonstrating the effectiveness of the proposed approach.https://isprs-annals.copernicus.org/articles/X-G-2025/47/2025/isprs-annals-X-G-2025-47-2025.pdf
spellingShingle R. Ali
M. Mehltretter
C. Heipke
Integrating Viewing Direction and Image Features for Robust Multi-View Multi-Object 3D Pedestrian Tracking
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
title Integrating Viewing Direction and Image Features for Robust Multi-View Multi-Object 3D Pedestrian Tracking
title_full Integrating Viewing Direction and Image Features for Robust Multi-View Multi-Object 3D Pedestrian Tracking
title_fullStr Integrating Viewing Direction and Image Features for Robust Multi-View Multi-Object 3D Pedestrian Tracking
title_full_unstemmed Integrating Viewing Direction and Image Features for Robust Multi-View Multi-Object 3D Pedestrian Tracking
title_short Integrating Viewing Direction and Image Features for Robust Multi-View Multi-Object 3D Pedestrian Tracking
title_sort integrating viewing direction and image features for robust multi view multi object 3d pedestrian tracking
url https://isprs-annals.copernicus.org/articles/X-G-2025/47/2025/isprs-annals-X-G-2025-47-2025.pdf
work_keys_str_mv AT rali integratingviewingdirectionandimagefeaturesforrobustmultiviewmultiobject3dpedestriantracking
AT mmehltretter integratingviewingdirectionandimagefeaturesforrobustmultiviewmultiobject3dpedestriantracking
AT cheipke integratingviewingdirectionandimagefeaturesforrobustmultiviewmultiobject3dpedestriantracking