Photogrammetry engaged automated image labeling approach

Deep learning models require many instances of training data to be able to accurately detect the desired object. However, the labeling of images is currently conducted manually due to the inclusion of irrelevant scenes in the original images, especially for the data collected in a dynamic environmen...

Full description

Saved in:
Bibliographic Details
Main Authors: Jonathan Boyack, Jongseong Brad Choi
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:Visual Informatics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2468502X25000221
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850183585558953984
author Jonathan Boyack
Jongseong Brad Choi
author_facet Jonathan Boyack
Jongseong Brad Choi
author_sort Jonathan Boyack
collection DOAJ
description Deep learning models require many instances of training data to be able to accurately detect the desired object. However, the labeling of images is currently conducted manually due to the inclusion of irrelevant scenes in the original images, especially for the data collected in a dynamic environment such as from drone imagery. In this work, we developed an automated extraction of training data set using photogrammetry. This approach works with continuous and arbitrary collection of visual data, such as video, encompassing a stationary object. A dense point cloud was first generated to estimate the geometric relationship between individual images using a structure-from-motion (SfM) technique, followed by user-designated region-of-interests, ROIs, that are automatically extracted from the original images. An orthophoto mosaic of the façade plane of the building shown in the point cloud was created to ease the user’s selection of an intended labeling region of the object, which is a one-time process. We verified this method by using the ROIs extracted from a previously obtained dataset to train and test a convolutional neural network which is modeled to detect damage locations. The method put forward in this work allows a relatively small amount of labeling to generate a large amount of training data. We successfully demonstrate the capabilities of the technique with the dataset previously collected by a drone from an abandoned building in which many of the glass windows have been damaged.
format Article
id doaj-art-fcc32e47e84f4a10a9f2e9b23f47c701
institution OA Journals
issn 2468-502X
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series Visual Informatics
spelling doaj-art-fcc32e47e84f4a10a9f2e9b23f47c7012025-08-20T02:17:19ZengElsevierVisual Informatics2468-502X2025-06-019210023910.1016/j.visinf.2025.100239Photogrammetry engaged automated image labeling approachJonathan Boyack0Jongseong Brad Choi1Department of Mechanical Engineering, SUNY Korea, Incheon, 21985, Republic of KoreaDepartment of Mechanical Engineering, SUNY Korea, Incheon, 21985, Republic of Korea; Department of Mechanical Engineering, Stony Brook University, Stony Brook, NY 11794, USA; Corresponding author at: Department of Mechanical Engineering, SUNY Korea, Incheon, 21985, Republic of Korea.Deep learning models require many instances of training data to be able to accurately detect the desired object. However, the labeling of images is currently conducted manually due to the inclusion of irrelevant scenes in the original images, especially for the data collected in a dynamic environment such as from drone imagery. In this work, we developed an automated extraction of training data set using photogrammetry. This approach works with continuous and arbitrary collection of visual data, such as video, encompassing a stationary object. A dense point cloud was first generated to estimate the geometric relationship between individual images using a structure-from-motion (SfM) technique, followed by user-designated region-of-interests, ROIs, that are automatically extracted from the original images. An orthophoto mosaic of the façade plane of the building shown in the point cloud was created to ease the user’s selection of an intended labeling region of the object, which is a one-time process. We verified this method by using the ROIs extracted from a previously obtained dataset to train and test a convolutional neural network which is modeled to detect damage locations. The method put forward in this work allows a relatively small amount of labeling to generate a large amount of training data. We successfully demonstrate the capabilities of the technique with the dataset previously collected by a drone from an abandoned building in which many of the glass windows have been damaged.http://www.sciencedirect.com/science/article/pii/S2468502X25000221PhotogrammetryDeep learningComputer visionStructure-from-motionOrthophotoROI
spellingShingle Jonathan Boyack
Jongseong Brad Choi
Photogrammetry engaged automated image labeling approach
Visual Informatics
Photogrammetry
Deep learning
Computer vision
Structure-from-motion
Orthophoto
ROI
title Photogrammetry engaged automated image labeling approach
title_full Photogrammetry engaged automated image labeling approach
title_fullStr Photogrammetry engaged automated image labeling approach
title_full_unstemmed Photogrammetry engaged automated image labeling approach
title_short Photogrammetry engaged automated image labeling approach
title_sort photogrammetry engaged automated image labeling approach
topic Photogrammetry
Deep learning
Computer vision
Structure-from-motion
Orthophoto
ROI
url http://www.sciencedirect.com/science/article/pii/S2468502X25000221
work_keys_str_mv AT jonathanboyack photogrammetryengagedautomatedimagelabelingapproach
AT jongseongbradchoi photogrammetryengagedautomatedimagelabelingapproach