All Issue

2026 Vol.14, Issue 1 Preview Page
Development of a YOLO-Based Deep Learning Algorithm For High-Precision Small Object Detection in Real-Time Hot Spot Analysis of Solar Panels

실시간 태양광 패널 핫스팟 분석을 위한 고정밀 소형 객체 탐지용 YOLO 기반의 딥러닝 알고리즘 개발

Seungho Woo1
,
Changjo Kim2*
,
Byeongsu Kim2*
우 승호1
,
김 창조2*
,
김 병수2*
1Department of Electrical, Electronic, and Computer Eengineering, University of Ulsan, Ulsan, 44610, Korea
2School of Electrical and Electronic Engineering, University of Ulsan, Ulsan, 44610, Korea
1전기전자정보시스템공학전공, 울산대학교, 울산, 44610
2전기전자융합학부, 울산대학교, 울산, 44610
*Corresponding Author
31 March 2026. pp. 38-47
PDF XML Citation
Abstract

With the expansion of solar energy as a leading renewable energy source due to its sustainability, ease of installation, and long operational lifetime, autonomous photovoltaic operation and maintenance (O&M) systems are increasingly important. Here, we present a YOLO_v11n-based deep learning model specializing in precise detection of hot spots on solar panels. We analyze the impact of augmentation hyperparameters (mixup, mosaic, copy_paste and close_mosaic) on small object detection. Our observations exhibit that a mixup value of 0.2 improves mAP@50, which represents the average accuracy, by over 3.7%, while excessive mixup increases image transparency, relaxing boundaries between small objects and reducing accuracy. Copy_paste has minimal effect when sufficient training data is available. Properly balancing the use of original and mosaic images, which combine four images into one, is crucial for high accuracy, and setting close_mosaic at 10% of entire training progress optimizes performance. Consequently, our model achieves over 95% of outstanding mAP@50, demonstrating robust and high-precision hot spot detection.

Keywords
Deep learning-based hot spot monitoring
Autonomous operation and maintenance
You only look once
Hyperparameter
High-precision small object detection
References
1

B. Parida, S. Iniyan, R. Goic, A review of solar photovoltaic technologies. Renew. Sustain. Energy Rev. 15(3), 1625-1636 (2011).

10.1016/j.rser.2010.11.032
2

P. A. Owusu, S. Asumadu-Sarkodie, A review of renewable energy sources, sustainability issues and climate change mitigation. Cogent Eng. 3(1), 1167990 (2016).

10.1080/23311916.2016.1167990
3

A. Züttel, N. Gallandat, P. J. Dyson, L. Schlapbach, P. W. Gilgen, S. I. Orimo, Future Swiss Energy Economy: the challenge of storing renewable energy. Front. Energy Res. 9, 785908 (2022).

10.3389/fenrg.2021.785908
4

International Energy Agency, “Renewables 2025: Analysis and forecast to 2030”, International Energy Agency, (2025).

5

Ministry of Trade, Industry and Resources, “The 11th Basic Plan for Long-Term Electricity Supply and Demand. (2024-2038)”, Ministry of Trade, Industry and Resources, (2025).

6

H. Abdulla, A. Sleptchenko, A. Nayfeh, Photovoltaic systems operation and maintenance: A review and future directions. Renew. Sustain. Energy Rev. 195, 114342 (2024).

10.1016/j.rser.2024.114342
7

K. A. K. Niazi, Y. Yang, D. Sera, Review of mismatch mitigation techniques for PV modules. IET Renew. Power Gener. 13(12), 2035-2050 (2019).

10.1049/iet-rpg.2019.0153
8

H. Al Mahdi, P. G. Leahy, M. Alghoul, A. P. Morrison, A review of photovoltaic module failure and degradation mechanisms: Causes and detection techniques. Solar 4(1), 43-82 (2024).

10.3390/solar4010003
9

Y. Wang, K. Itako, T. Kudoh, K. Koh, Q. Ge, Voltage-based hot-spot detection method for photovoltaic string using a projector. Energies 10(2), 230 (2017).

10.3390/en10020230
10

N. Kumari, S. K. Singh, S. Kumar, V. K. Jadoun, Performance analysis of partially shaded high-efficiency mono PERC/mono crystalline PV module under indoor and environmental conditions. Sci. Rep. 14, 21587 (2024).

10.1038/s41598-024-72502-z39285257PMC11405760
11

S. Ghosh, S. K. Singh, V. K. Yadav, Experimental investigation of hotspot phenomenon in PV arrays under mismatch conditions. Sol. Energy 253, 219-230 (2023).

10.1016/j.solener.2023.02.033
12

M. Dhimish, A. M. Tyrrell, Power loss and hotspot analysis for photovoltaic modules affected by potential induced degradation. npj Mater. Degrad. 6(1), 11 (2022).

10.1038/s41529-022-00221-9
13

M. Dhimish, P. Mather, V. Holmes, Evaluating power loss and performance ratio of hot-spotted photovoltaic modules. IEEE Trans. Electron Devices 65(12), 5419-5427 (2018).

10.1109/TED.2018.2877806
14

A. H. Numan, H. A. Hussein, Z. S. Dawood, Hot spot analysis of photovoltaic module under partial shading conditions by using IR-imaging technology. Eng. Technol. J. 39(9), 1338-1344 (2021).

10.30684/etj.v39i9.841
15

G. D. Lee, D. O. Im, “Establishment and Operation of Long-Term LCOE Forecast System for Expansion of Renewable Energy(5/5)”, Korea Energy Economics Institute, (2024).

16

T. Tajwar, O. H. Mobin, F. R. Khan, S. F. Hossain, M. Islam, M. M. Rahman, Infrared thermography based hotspot detection of photovoltaic module using YOLO. In 2021 IEEE 12th Energy Conversion Congress & Exposition-Asia (ECCE-Asia) (pp. 1542-1547) (2021).

10.1109/ECCE-Asia49820.2021.9478998
17

X. Han, X. Wang, C. Chen, G. Li, C. Piao, Hot spot detection of thermal infrared image of photovoltaic power station based on multi-task fusion. J. Inf. Process. Syst. 19(6), 791-802 (2023).

10.3745/JIPS.02.0208
18

M. U. Ali, H. F. Khan, M. Masud, K. D. Kallu, A. Zafar, A machine learning framework to identify the hotspot in photovoltaic module using infrared thermography. Sol. Energy 208, 643-651 (2020).

10.1016/j.solener.2020.08.027
19

S. Ren, K. He, R. Girshick, J. Sun, Faster R-CNN: Towards real-time object detection with region proposal networks. IEEE Trans. Pattern Anal. Mach. Intell. 39(6), 1137-1149 (2016).

10.1109/TPAMI.2016.2577031
20

F. Joiya, Object detection: yolo vs faster R-CNN. Int. Res. J. Mod. Eng. Technol. Sci. 9, 1911-1915 (2022).

21

B. Karbouja, G. A. Topalian-Rivas, J. Krüger, Comparative performance evaluation of one-stage and two-stage object detectors for screw head detection and classification in disassembly processes. Procedia CIRP 122, 527-532 (2024).

10.1016/j.procir.2024.01.077
22

N. Jegham, C. Y. Koh, M. Abdelatti, A. Hendawi, Yolo evolution: A comprehensive benchmark and architectural review of yolov12, yolo11, and their previous versions. arXiv preprint arXiv:2411.00201 (2024).

10.2139/ssrn.5175639
23

A. A. Murat, M. S. Kiran, A comprehensive review on YOLO versions for object detection. Eng. Sci. Technol. Int. J. 70, 102161 (2025).

10.1016/j.jestch.2025.102161
24

R. Khanam, T. Asghar, M. Hussain, Comparative performance evaluation of yolov5, yolov8, and yolov11 for solar panel defect detection. Solar 5(1), 1-25 (2025).

10.3390/solar5010006
25

W. Cai, H. Lv, YOLO-PV: An enhanced YOLO11n model with multi-scale feature fusion for photovoltaic panel defect detection. Energies 18(20), 5489 (2025).

10.3390/en18205489
26

M. B. Çetinkaya, M. Gürek, Genetic algorithm-based hyperparameter optimization of YOLOv11: Real-time defect detection and classification in energy cable production lines. Arab. J. Sci. Eng. 1-15 (2025).

10.1007/s13369-025-10521-2
27

H. Zhang, M. Cisse, Y. N. Dauphin, D. Lopez-Paz, mixup: Beyond Empirical Risk Minimization. arXiv preprint arXiv:1710.09412 (2017).

28

G. Ghiasi, Y. Cui, A. Srinivas, R. Qian, T.-Y. Lin, E. D. Cubuk, Q. V. Le, B. Zoph, Simple copy-paste is a strong data augmentation method for instance segmentation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 2918-2928) (2021).

10.1109/CVPR46437.2021.00294
29

M. A. Adam, J. R. Tapamo, Enhancing YOLOv5 for autonomous driving: Efficient attention-based object detection on edge devices. J. Imaging 11(8), 263 (2025).

10.3390/jimaging1108026340863473PMC12387144
Information
  • Publisher :Korea Photovoltaic Society
  • Publisher(Ko) :한국태양광발전학회
  • Journal Title :Current Photovoltaic Research
  • Volume : 14
  • No :1
  • Pages :38-47
  • Received Date : 2025-12-03
  • Revised Date : 2026-01-03
  • Accepted Date : 2026-01-09
  • DOI :https://doi.org/10.21218/CPR.2026.14.1.038
Journal Informaiton Current Photovoltaic Research Current Photovoltaic Research
Online Submission
  • NRF
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close