Hayley Hames
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Hayley Hames has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
Cade Wagner
Department of Mathematical Sciences, University of Arkansas, Fayetteville, Arkansas, USA
1 shared publication- Cameron Andrew Petrie
University of Cambridge
1 shared publication Carla Klehm
1 shared publication- Geoffery Agorku
Department of Civil Engineering, University of Arkansas, Fayetteville, Arkansas, USA
1 shared publication - Hank Theiss
National Geospatial-Intelligence Agency (NGA) Sensor Geopositioning Center - Contractor, 7500 GEOINT Drive, MS S73-IBR, Springfield, Virginia 22150 USA
1 shared publication Jackson Cothren
1 shared publication
Latest Publications
- 29 October 2026
- Published by SAGE Publications in Transportation Research Record: Journal of the Transportation Research Board
Enhancing Maritime Domain Awareness on Inland Waterways: A You Only Look Once-Based Fusion of Satellite and Automatic Identification System for Vessel Characterization
Maritime domain awareness (MDA) in inland waterways is constrained by vulnerabilities in cooperative monitoring systems. This study presents a data fusion framework that integrates high-resolution satellite imagery with vessel trajectory data from the automatic identification system (AIS) to enhance detection, characterization, and tracking of maritime activity. The proposed approach leverages noncooperative satellite sensing to complement AIS data, enabling the identification of dark vessels, validation of cooperative traffic, and support for advanced MDA capabilities. A You Only Look Once (YOLO) v11-based object detection model is developed to detect and characterize vessels and barges by type, cover status, operational state, barge count, and direction of movement. The model is trained on a manually annotated dataset comprising 4,550 instances extracted from 5,973 mi 2 of Lower Mississippi River imagery. Evaluation on a held-out test set demonstrates strong performance across multiple tasks, with vessel classification achieving an F1-score of 95.8%, barge cover detection attaining an F1-score of 91.6%, and operational status classification reaching an F1-score of 99.4%. Direction-of-movement classification attains an accuracy of 98.1%, while barge-count estimation yields a mean absolute error of 2.4 barges. Spatial transferability assessment across a geographically distinct river system indicates minimal performance degradation, with accuracy levels remaining as high as 98%. These results demonstrate the effectiveness of integrating satellite imagery with AIS data for inland waterway monitoring. The framework supports near-real-time fleet inventory generation, anomaly detection, and improved situational awareness. Future work will focus on expanding dataset diversity, incorporating temporal modeling, and exploring multimodal learning to further enhance scalability and operational deployment.
- 29 October 2026
- Published by SAGE Publications in Transportation Research Record: Journal of the Transportation Research Board
Enhancing Maritime Domain Awareness on Inland Waterways: A You Only Look Once-Based Fusion of Satellite and Automatic Identification System for Vessel Characterization
Maritime domain awareness (MDA) in inland waterways is constrained by vulnerabilities in cooperative monitoring systems. This study presents a data fusion framework that integrates high-resolution satellite imagery with vessel trajectory data from the automatic identification system (AIS) to enhance detection, characterization, and tracking of maritime activity. The proposed approach leverages noncooperative satellite sensing to complement AIS data, enabling the identification of dark vessels, validation of cooperative traffic, and support for advanced MDA capabilities. A You Only Look Once (YOLO) v11-based object detection model is developed to detect and characterize vessels and barges by type, cover status, operational state, barge count, and direction of movement. The model is trained on a manually annotated dataset comprising 4,550 instances extracted from 5,973 mi 2 of Lower Mississippi River imagery. Evaluation on a held-out test set demonstrates strong performance across multiple tasks, with vessel classification achieving an F1-score of 95.8%, barge cover detection attaining an F1-score of 91.6%, and operational status classification reaching an F1-score of 99.4%. Direction-of-movement classification attains an accuracy of 98.1%, while barge-count estimation yields a mean absolute error of 2.4 barges. Spatial transferability assessment across a geographically distinct river system indicates minimal performance degradation, with accuracy levels remaining as high as 98%. These results demonstrate the effectiveness of integrating satellite imagery with AIS data for inland waterway monitoring. The framework supports near-real-time fleet inventory generation, anomaly detection, and improved situational awareness. Future work will focus on expanding dataset diversity, incorporating temporal modeling, and exploring multimodal learning to further enhance scalability and operational deployment.
- Open Access
- 2 September 2026
- Published by Elsevier in SSRN Electronic Journal
Using Declassified Imagery: Issues and Approaches
The declassification of Cold War-era, US spy satellite imagery, collectively known as the Keyhole imagery, has contributed to hundreds of research projects across disciplines including geography, ecology, glaciology, archaeology, heritage studies, and urban studies. Corona (KH-4) and, increasingly, Hexagon (KH-9) imagery has garnered attention and excitement throughout the scientific community. The worldwide, repeat coverage and high resolution (up to 0.6m) of this imagery, which also has stereo coverage, is exceptional and not replicated elsewhere; it is also low-cost. However, declassified imagery has complex geometries and issues associated with processing and modeling. Even the initial imagery selection can deter adoption. This article addresses the critical issues involved with manipulating these declassified imagery. It synthesizes dozens of case studies across scientific domains to draw out commonalities in experiences at the selection, processing, and modeling stages. Issues with the image scans and ordering process, over- and underavailability of imagery, mosaicking, lack of metadata, ground control points, rigorous modeling, and image size are outlined. Approaches help guide researchers towards tailored solutions. Approaches to efficiently and effectively sort through imagery archives and georeferencing software options, and data management strategies are discussed. Two open-access tools for modeling, Sunspot and Ames Stereo Pipeline, are highlighted. The aim throughout is to increase the use of declassified historical imagery by orienting researchers to issues they can anticipate and approaches they can employ. The article represents the most comprehensive discussion of methods associated with declassified imagery published to date, that will enable accessibility and innovation.
- 2 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessUsing Declassified Imagery: Issues and Approaches
The declassification of Cold War-era, US spy satellite imagery, collectively known as the Keyhole imagery, has contributed to hundreds of research projects across disciplines including geography, ecology, glaciology, archaeology, heritage studies, and urban studies. Corona (KH-4) and, increasingly, Hexagon (KH-9) imagery has garnered attention and excitement throughout the scientific community. The worldwide, repeat coverage and high resolution (up to 0.6m) of this imagery, which also has stereo coverage, is exceptional and not replicated elsewhere; it is also low-cost. However, declassified imagery has complex geometries and issues associated with processing and modeling. Even the initial imagery selection can deter adoption. This article addresses the critical issues involved with manipulating these declassified imagery. It synthesizes dozens of case studies across scientific domains to draw out commonalities in experiences at the selection, processing, and modeling stages. Issues with the image scans and ordering process, over- and underavailability of imagery, mosaicking, lack of metadata, ground control points, rigorous modeling, and image size are outlined. Approaches help guide researchers towards tailored solutions. Approaches to efficiently and effectively sort through imagery archives and georeferencing software options, and data management strategies are discussed. Two open-access tools for modeling, Sunspot and Ames Stereo Pipeline, are highlighted. The aim throughout is to increase the use of declassified historical imagery by orienting researchers to issues they can anticipate and approaches they can employ. The article represents the most comprehensive discussion of methods associated with declassified imagery published to date, that will enable accessibility and innovation.