From Simulation to Reality: Transfer Learning for Automating Pseudo‐Labeling of Real and Infrared Imagery

Training a convolutional neural network (CNN) for real‐world applications is challenging due to the requirement of high‐quality labeled imagery. This study employs pseudo‐labeling and transfer learning, built upon a 6D pose estimation framework. A CNN trained on synthetic images predicts bounding bo...

Full description

Saved in:
Bibliographic Details
Main Authors: Jeffrey Choate, Derek Worth, Scott L. Nykl, Clark Taylor, Brett Borghetti, Christine Schubert Kabban, Ryan Raettig
Format: Article
Language:English
Published: Wiley 2025-05-01
Series:Advanced Intelligent Systems
Subjects:
Online Access:https://doi.org/10.1002/aisy.202400575
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850129523061817344
author Jeffrey Choate
Derek Worth
Scott L. Nykl
Clark Taylor
Brett Borghetti
Christine Schubert Kabban
Ryan Raettig
author_facet Jeffrey Choate
Derek Worth
Scott L. Nykl
Clark Taylor
Brett Borghetti
Christine Schubert Kabban
Ryan Raettig
author_sort Jeffrey Choate
collection DOAJ
description Training a convolutional neural network (CNN) for real‐world applications is challenging due to the requirement of high‐quality labeled imagery. This study employs pseudo‐labeling and transfer learning, built upon a 6D pose estimation framework. A CNN trained on synthetic images predicts bounding boxes (bbox) for an object's components in a real image. With as few as four bbox predictions, the framework solves for the object's pose relative to the camera and reprojects bboxes for all components onto that image. The pose and reprojections allow filtering of bad predictions, a common issue in pseudo‐labeling. Thereby, enabling automated labeling of large datasets with minimal human intervention. Tested on color and long‐wave infrared imagery captured during December 2023 flight tests, this process demonstrates increased predictions, enhanced performance across situations, reduced reprojection error, and stabilized pose predictions. This technique is significant as it enables labeling of real‐world imagery without expensive truth systems, requiring only a camera. It supports learning and labeling of previously captured imagery without known camera calibrations, facilitating labeled data creation for impractical‐to‐simulate sensors. Ultimately, this transfer learning approach provides a low‐cost and precise method for creating CNNs trained on operationally relevant data, previously unattainable by the everyday user.
format Article
id doaj-art-b2623a7edfd94a1e8dfcdcd9cbfb84db
institution OA Journals
issn 2640-4567
language English
publishDate 2025-05-01
publisher Wiley
record_format Article
series Advanced Intelligent Systems
spelling doaj-art-b2623a7edfd94a1e8dfcdcd9cbfb84db2025-08-20T02:32:56ZengWileyAdvanced Intelligent Systems2640-45672025-05-0175n/an/a10.1002/aisy.202400575From Simulation to Reality: Transfer Learning for Automating Pseudo‐Labeling of Real and Infrared ImageryJeffrey Choate0Derek Worth1Scott L. Nykl2Clark Taylor3Brett Borghetti4Christine Schubert Kabban5Ryan Raettig6Department of Electrical and Computer Engineering Air Force Institute of Technology Wright‐Patterson AFB OH 45433 USADepartment of Electrical and Computer Engineering Air Force Institute of Technology Wright‐Patterson AFB OH 45433 USADepartment of Electrical and Computer Engineering Air Force Institute of Technology Wright‐Patterson AFB OH 45433 USADepartment of Electrical and Computer Engineering Air Force Institute of Technology Wright‐Patterson AFB OH 45433 USADepartment of Electrical and Computer Engineering Air Force Institute of Technology Wright‐Patterson AFB OH 45433 USADepartment of Mathematics and Statistics Air Force Institute of Technology Wright‐Patterson AFB OH 45433 USADepartment of Electrical and Computer Engineering Air Force Institute of Technology Wright‐Patterson AFB OH 45433 USATraining a convolutional neural network (CNN) for real‐world applications is challenging due to the requirement of high‐quality labeled imagery. This study employs pseudo‐labeling and transfer learning, built upon a 6D pose estimation framework. A CNN trained on synthetic images predicts bounding boxes (bbox) for an object's components in a real image. With as few as four bbox predictions, the framework solves for the object's pose relative to the camera and reprojects bboxes for all components onto that image. The pose and reprojections allow filtering of bad predictions, a common issue in pseudo‐labeling. Thereby, enabling automated labeling of large datasets with minimal human intervention. Tested on color and long‐wave infrared imagery captured during December 2023 flight tests, this process demonstrates increased predictions, enhanced performance across situations, reduced reprojection error, and stabilized pose predictions. This technique is significant as it enables labeling of real‐world imagery without expensive truth systems, requiring only a camera. It supports learning and labeling of previously captured imagery without known camera calibrations, facilitating labeled data creation for impractical‐to‐simulate sensors. Ultimately, this transfer learning approach provides a low‐cost and precise method for creating CNNs trained on operationally relevant data, previously unattainable by the everyday user.https://doi.org/10.1002/aisy.202400575convolutional neural networkPerspective‐N‐Point algorithmpseudo‐labelingreal time pose estimationsynthetic imagery generationtransfer learning
spellingShingle Jeffrey Choate
Derek Worth
Scott L. Nykl
Clark Taylor
Brett Borghetti
Christine Schubert Kabban
Ryan Raettig
From Simulation to Reality: Transfer Learning for Automating Pseudo‐Labeling of Real and Infrared Imagery
Advanced Intelligent Systems
convolutional neural network
Perspective‐N‐Point algorithm
pseudo‐labeling
real time pose estimation
synthetic imagery generation
transfer learning
title From Simulation to Reality: Transfer Learning for Automating Pseudo‐Labeling of Real and Infrared Imagery
title_full From Simulation to Reality: Transfer Learning for Automating Pseudo‐Labeling of Real and Infrared Imagery
title_fullStr From Simulation to Reality: Transfer Learning for Automating Pseudo‐Labeling of Real and Infrared Imagery
title_full_unstemmed From Simulation to Reality: Transfer Learning for Automating Pseudo‐Labeling of Real and Infrared Imagery
title_short From Simulation to Reality: Transfer Learning for Automating Pseudo‐Labeling of Real and Infrared Imagery
title_sort from simulation to reality transfer learning for automating pseudo labeling of real and infrared imagery
topic convolutional neural network
Perspective‐N‐Point algorithm
pseudo‐labeling
real time pose estimation
synthetic imagery generation
transfer learning
url https://doi.org/10.1002/aisy.202400575
work_keys_str_mv AT jeffreychoate fromsimulationtorealitytransferlearningforautomatingpseudolabelingofrealandinfraredimagery
AT derekworth fromsimulationtorealitytransferlearningforautomatingpseudolabelingofrealandinfraredimagery
AT scottlnykl fromsimulationtorealitytransferlearningforautomatingpseudolabelingofrealandinfraredimagery
AT clarktaylor fromsimulationtorealitytransferlearningforautomatingpseudolabelingofrealandinfraredimagery
AT brettborghetti fromsimulationtorealitytransferlearningforautomatingpseudolabelingofrealandinfraredimagery
AT christineschubertkabban fromsimulationtorealitytransferlearningforautomatingpseudolabelingofrealandinfraredimagery
AT ryanraettig fromsimulationtorealitytransferlearningforautomatingpseudolabelingofrealandinfraredimagery