Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI)

Oscar Best; Asiya Khan; Mario Gianni; Sanjay Sharma; Keri Collins

doi:10.1007/978-3-031-47724-9_37

Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI)

Oscar Best^*
, Asiya Khan
, Mario Gianni
, Sanjay Sharma
, Keri Collins

^*Corresponding author for this work

University of Liverpool

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings published in a book › peer-review

Abstract

This paper expands on the publicly available dataset of wind turbine surface damage and uses this dataset to finetune all P5 models for both YOLOv5 and YOLOv7 object detection frameworks. YOLOv7 outperformed YOLOv5, with the YOLOv7x model achieving the highest recall score and best F1-confidence. This model was therefore selected for inference on both images and video of wind turbines. Slicing Aided Hyper Inference (SAHI) has also been used to improve detection capability for smaller instances of damage. The model was further evaluated on a dataset collected from a scaled model of a wind turbine, with hand drawn damages. Lastly, this dataset was used for inference using SAHI, which showed slight improvement for detecting damage instances. More accurate results were observed when evaluating the model on real damage examples compared to simulated damage.

Original language	English
Title of host publication	Intelligent Systems and Applications - Proceedings of the 2023 Intelligent Systems Conference IntelliSys Volume 2
Editors	Kohei Arai
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	564-576
Number of pages	13
ISBN (Print)	9783031477232
DOIs	https://doi.org/10.1007/978-3-031-47724-9_37
Publication status	Published - 2024
Event	Intelligent Systems Conference, IntelliSys 2023 - Amsterdam, Netherlands Duration: 7 Sept 2023 → 8 Sept 2023

Publication series

Name	Lecture Notes in Networks and Systems
Volume	823 LNNS
ISSN (Print)	2367-3370
ISSN (Electronic)	2367-3389

Conference

Conference	Intelligent Systems Conference, IntelliSys 2023
Country/Territory	Netherlands
City	Amsterdam
Period	7/09/23 → 8/09/23

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

ASJC Scopus subject areas

Control and Systems Engineering
Signal Processing
Computer Networks and Communications

Keywords

Computer Vision
Machine Learning
Wind Turbine Inspection

Access to Document

10.1007/978-3-031-47724-9_37

Cite this

Best, O., Khan, A., Gianni, M., Sharma, S., & Collins, K. (2024). Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI). In K. Arai (Ed.), Intelligent Systems and Applications - Proceedings of the 2023 Intelligent Systems Conference IntelliSys Volume 2 (pp. 564-576). (Lecture Notes in Networks and Systems; Vol. 823 LNNS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-47724-9_37

Best, Oscar ; Khan, Asiya ; Gianni, Mario et al. / Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI). Intelligent Systems and Applications - Proceedings of the 2023 Intelligent Systems Conference IntelliSys Volume 2. editor / Kohei Arai. Springer Science and Business Media Deutschland GmbH, 2024. pp. 564-576 (Lecture Notes in Networks and Systems).

@inproceedings{d34caec0d644408993108e3bfb1544c6,

title = "Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI)",

abstract = "This paper expands on the publicly available dataset of wind turbine surface damage and uses this dataset to finetune all P5 models for both YOLOv5 and YOLOv7 object detection frameworks. YOLOv7 outperformed YOLOv5, with the YOLOv7x model achieving the highest recall score and best F1-confidence. This model was therefore selected for inference on both images and video of wind turbines. Slicing Aided Hyper Inference (SAHI) has also been used to improve detection capability for smaller instances of damage. The model was further evaluated on a dataset collected from a scaled model of a wind turbine, with hand drawn damages. Lastly, this dataset was used for inference using SAHI, which showed slight improvement for detecting damage instances. More accurate results were observed when evaluating the model on real damage examples compared to simulated damage.",

keywords = "Computer Vision, Machine Learning, Wind Turbine Inspection",

author = "Oscar Best and Asiya Khan and Mario Gianni and Sanjay Sharma and Keri Collins",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.; Intelligent Systems Conference, IntelliSys 2023 ; Conference date: 07-09-2023 Through 08-09-2023",

year = "2024",

doi = "10.1007/978-3-031-47724-9\_37",

language = "English",

isbn = "9783031477232",

series = "Lecture Notes in Networks and Systems",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "564--576",

editor = "Kohei Arai",

booktitle = "Intelligent Systems and Applications - Proceedings of the 2023 Intelligent Systems Conference IntelliSys Volume 2",

address = "Germany",

}

Best, O , Khan, A, Gianni, M, Sharma, S & Collins, K 2024, Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI). in K Arai (ed.), Intelligent Systems and Applications - Proceedings of the 2023 Intelligent Systems Conference IntelliSys Volume 2. Lecture Notes in Networks and Systems, vol. 823 LNNS, Springer Science and Business Media Deutschland GmbH, pp. 564-576, Intelligent Systems Conference, IntelliSys 2023, Amsterdam, Netherlands, 7/09/23. https://doi.org/10.1007/978-3-031-47724-9_37

Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI). / Best, Oscar ; Khan, Asiya; Gianni, Mario et al.
Intelligent Systems and Applications - Proceedings of the 2023 Intelligent Systems Conference IntelliSys Volume 2. ed. / Kohei Arai. Springer Science and Business Media Deutschland GmbH, 2024. p. 564-576 (Lecture Notes in Networks and Systems; Vol. 823 LNNS).

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings published in a book › peer-review

TY - GEN

T1 - Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI)

AU - Best, Oscar

AU - Khan, Asiya

AU - Gianni, Mario

AU - Sharma, Sanjay

AU - Collins, Keri

PY - 2024

Y1 - 2024

N2 - This paper expands on the publicly available dataset of wind turbine surface damage and uses this dataset to finetune all P5 models for both YOLOv5 and YOLOv7 object detection frameworks. YOLOv7 outperformed YOLOv5, with the YOLOv7x model achieving the highest recall score and best F1-confidence. This model was therefore selected for inference on both images and video of wind turbines. Slicing Aided Hyper Inference (SAHI) has also been used to improve detection capability for smaller instances of damage. The model was further evaluated on a dataset collected from a scaled model of a wind turbine, with hand drawn damages. Lastly, this dataset was used for inference using SAHI, which showed slight improvement for detecting damage instances. More accurate results were observed when evaluating the model on real damage examples compared to simulated damage.

AB - This paper expands on the publicly available dataset of wind turbine surface damage and uses this dataset to finetune all P5 models for both YOLOv5 and YOLOv7 object detection frameworks. YOLOv7 outperformed YOLOv5, with the YOLOv7x model achieving the highest recall score and best F1-confidence. This model was therefore selected for inference on both images and video of wind turbines. Slicing Aided Hyper Inference (SAHI) has also been used to improve detection capability for smaller instances of damage. The model was further evaluated on a dataset collected from a scaled model of a wind turbine, with hand drawn damages. Lastly, this dataset was used for inference using SAHI, which showed slight improvement for detecting damage instances. More accurate results were observed when evaluating the model on real damage examples compared to simulated damage.

KW - Computer Vision

KW - Machine Learning

KW - Wind Turbine Inspection

UR - https://www.scopus.com/pages/publications/85192155448

U2 - 10.1007/978-3-031-47724-9_37

DO - 10.1007/978-3-031-47724-9_37

M3 - Conference proceedings published in a book

AN - SCOPUS:85192155448

SN - 9783031477232

T3 - Lecture Notes in Networks and Systems

SP - 564

EP - 576

BT - Intelligent Systems and Applications - Proceedings of the 2023 Intelligent Systems Conference IntelliSys Volume 2

A2 - Arai, Kohei

PB - Springer Science and Business Media Deutschland GmbH

T2 - Intelligent Systems Conference, IntelliSys 2023

Y2 - 7 September 2023 through 8 September 2023

ER -

Best O , Khan A, Gianni M, Sharma S, Collins K. Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI). In Arai K, editor, Intelligent Systems and Applications - Proceedings of the 2023 Intelligent Systems Conference IntelliSys Volume 2. Springer Science and Business Media Deutschland GmbH. 2024. p. 564-576. (Lecture Notes in Networks and Systems). doi: 10.1007/978-3-031-47724-9_37

Wind Turbine Surface Damage Detection Using YOLOv7 with Slicing Aided Hyper Inference (SAHI)

Abstract

Publication series

Conference

UN SDGs

ASJC Scopus subject areas

Keywords

Access to Document

Other files and links

Fingerprint

Cite this