Korean Journal of Agricultural and Forest Meteorology (한국농림기상학회지)

Korean Society of Agricultural and Forest Meteorology (한국농림기상학회)
권호 표지
DOI QR코드

DOI QR Code

Sorghum Panicle Detection using YOLOv5 based on RGB Image Acquired by UAV System

무인기로 취득한 RGB 영상과 YOLOv5를 이용한 수수 이삭 탐지

  • Min-Jun, Park (Department of Bio-System Engineering, Gyengsang National University (Institute of Agriculture & Life Science)) ;
  • Chan-Seok, Ryu (Department of Bio-System Engineering, Gyengsang National University (Institute of Agriculture & Life Science)) ;
  • Ye-Seong, Kang (Department of Bio-System Engineering, Gyengsang National University (Institute of Agriculture & Life Science)) ;
  • Hye-Young, Song (Department of Bio-System Engineering, Gyengsang National University (Institute of Agriculture & Life Science)) ;
  • Hyun-Chan, Baek (Department of Bio-System Engineering, Gyengsang National University (Institute of Agriculture & Life Science)) ;
  • Ki-Su, Park (Department of Bio-System Engineering, Gyengsang National University (Institute of Agriculture & Life Science)) ;
  • Eun-Ri, Kim (Department of Bio-System Engineering, Gyengsang National University (Institute of Agriculture & Life Science)) ;
  • Jin-Ki, Park (Southern Crop Department, National Institute of Crop Science, Rural Development Administration) ;
  • Si-Hyeong, Jang (Fruit Research Division, National institute of Horticultural & Herbal Science)
  • 박민준 (경상국립대학교 바이오시스템공학과 (농업생명과학연구원)) ;
  • 유찬석 (경상국립대학교 바이오시스템공학과 (농업생명과학연구원)) ;
  • 강예성 (경상국립대학교 바이오시스템공학과 (농업생명과학연구원)) ;
  • 송혜영 (경상국립대학교 바이오시스템공학과 (농업생명과학연구원)) ;
  • 백현찬 (경상국립대학교 바이오시스템공학과 (농업생명과학연구원)) ;
  • 박기수 (경상국립대학교 바이오시스템공학과 (농업생명과학연구원)) ;
  • 김은리 (경상국립대학교 바이오시스템공학과 (농업생명과학연구원)) ;
  • 박진기 (국립식량과학원 남부작물부 생산기술개발과) ;
  • 장시형 (국립원예특작과학원 원예작물부 과수과)
  • Received : 2022.12.07
  • Accepted : 2022.12.20
  • Published : 2022.12.30
Download PDF
⟨ Previous Next ⟩

Abstract

The purpose of this study is to detect the sorghum panicle using YOLOv5 based on RGB images acquired by a unmanned aerial vehicle (UAV) system. The high-resolution images acquired using the RGB camera mounted in the UAV on September 2, 2022 were split into 512×512 size for YOLOv5 analysis. Sorghum panicles were labeled as bounding boxes in the split image. 2,000images of 512×512 size were divided at a ratio of 6:2:2 and used to train, validate, and test the YOLOv5 model, respectively. When learning with YOLOv5s, which has the fewest parameters among YOLOv5 models, sorghum panicles were detected with mAP@50=0.845. In YOLOv5m with more parameters, sorghum panicles could be detected with mAP@50=0.844. Although the performance of the two models is similar, YOLOv5s ( 4 hours 35 minutes) has a faster training time than YOLOv5m (5 hours 15 minutes). Therefore, in terms of time cost, developing the YOLOv5s model was considered more efficient for detecting sorghum panicles. As an important step in predicting sorghum yield, a technique for detecting sorghum panicles using high-resolution RGB images and the YOLOv5 model was presented.

본 연구는 수수의 수확량 추정을 위해 무인기로 취득한 RGB 영상과 YOLOv5를 이용하여 수수 이삭 탐지 모델을 개발하였다. 이삭이 가장 잘 식별되는 9월 2일의 영상 중 512×512로 분할된 2000장을 이용하여 모델의 학습, 검증 및 테스트하였다. YOLOv5의 모델 중 가장 파라미터가 적은 YOLOv5s에서 mAP@50=0.845로 수수 이삭을 탐지할 수 있었다. 파라미터가 증가한 YOLOv5m에서는 mAP@50=0.844로 수수 이삭을 탐지할 수 있었다. 두 모델의 성능이 유사하나 YOLOv5s (4시간 35분)가 YOLOv5m (5시간 15분)보다 훈련시간이 더 빨라 YOLOv5s가 수수 이삭 탐지에 효율적이라고 판단된다. 개발된 모델을 이용하여 수수의 수확량 예측을 위한 단위면적당 이삭 수를 추정하는 알고리즘의 기초자료로 유용하게 활용될 것으로 판단된다. 추가적으로 아직 개발의 초기 단계를 감안하면 확보된 데이터를 이용하여 성능 개선 및 다른 CNN 모델과 비교 검토할 필요가 있다고 사료된다.

Keywords

Acknowledgement

본 성과물은 농촌진흥청 연구사업(세부과제번호: PJ0157532022)의 지원에 의해 수행되었음.

References

  1. Alzubaidi, L., J. Zhang, A. J. Humaidi, A. Al-Dujaili, Y. Duan, O. Al-Shamma, J. Santamaria, M. A. Fadhel, M. Al-Amidie, and L. Farhan, 2021: Review of deep learning: Concepts, CNN architectures, challenges, applications, future directions. Journal of big Data 8(1), 1-74. https://doi.org/10.1186/s40537-020-00387-6
  2. Bayraktar, E., M. E. Basarkan, and N. Celebi, 2020: A low-cost UAV framework towards ornamental plant detection and counting in the wild. ISPRS Journal of Photogrammetry and Remote Sensing 167, 1-11. https://doi.org/10.1016/j.isprsjprs.2020.06.012
  3. David, E., G. Daubige, F. Joudelat, P. Burger, A. Comar, B. de Solan, and F. Baret, 2022: Plant detection and counting from high-resolution RGB images acquired from UAVs: comparison between deep-learning and handcrafted methods with application to maize, sugar beet, and sunflower. bioRxiv, 2021-04.
  4. Du, Z., J. Yang, C. Ou, and T. Zhang, 2019: Smallholder crop area mapped with a semantic segmentation deep learning method. Remote Sensing 11(7), 888. https://doi.org/10.3390/rs11070888
  5. Garcia-Martinez, H., H. Flores-Magdaleno, A. Khalil-Gardezi, R. Ascencio-Hernandez, L. Tijerina-Chavez, M. A. Vazquez-Pena, and O. R. Mancilla-Villa, 2020: Digital count of corn plants using images taken by unmanned aerial vehicles and cross correlation of templates. Agronomy 10(4), 469. https://doi.org/10.3390/agronomy10040469
  6. Gonzalo-Martin, C., A. Garcia-Pedrero, and M. Lillo-Saavedra, 2021: Improving deep learning sorghum head detection through test time augmentation. Computers and Electronics in Agriculture 186, 106179. https://doi.org/10.1016/j.compag.2021.106179
  7. Ha, Y. D., and S. P. Lee, 2001: Characteristics of proteins in Italian millet, sorghum and common millet. Korean J Postharvest Sci Technol 8(2), 187-192.
  8. NICS (National Institute of Crop Science), 2016: http://www.nics.go.kr/bbs/view.do?m=100000126&bbsId=research&bbsSn=202912
  9. NICS (National Institute of Crop Science), 2019: http://www.nics.go.kr/bbs/view.do?m=100000126&bbsId=research&bbsSn=463112
  10. KMA (Korea Meteorological Administration), 2022: https://data.kma.go.kr/data/grnd/selectAsosRltmList.do?pgmNo=36
  11. KOSIS (Korea Statistical Information Service), 2022: Grain Production, https://kosis.kr/
  12. Li, H., P. Wang, and C. Huang, 2022: Comparison of Deep Learning Methods for Detecting and Counting Sorghum Heads in UAV Imagery. Remote Sensing 14(13), 3143. https://doi.org/10.3390/rs14133143
  13. Marsalis, M. A., S. V. Angadi, and F. E. Contreras-Govea, 2010: Dry matter yield and nutritive value of corn, forage sorghum, and BMR forage sorghum at different plant populations and nitrogen rates. Field Crops Research 116(1-2), 52-57. https://doi.org/10.1016/j.fcr.2009.11.009
  14. MAFRA (Ministry of Agriculture, Food and Rural Affairs of Korea), 2021: 양정자료. 36-37, 11-1543000-000002-10
  15. Moon, H., J. H. Ryu, S. I. Na, S. W. Jang, S. H. Sin, and J. Cho, 2021: Comparative Analysis of Rice Lodging Area Using a UAV-based Multispectral Imagery. Korean Journal of Remote Sensing 37(5_1), 917-926. https://doi.org/10.7780/KJRS.2021.37.5.1.7
  16. Na, S., C. Park, K. So, H. Ahn, and K. Lee, 2018: Application method of unmanned aerial vehicle for crop monitoring in Korea. Korean Journal of Remote Sensing 34(5), 829-846. https://doi.org/10.7780/KJRS.2018.34.5.10
  17. Prasad, P. V., and S. A. Staggenborg, 2009: Growth and production of sorghum and millets. Soils, plant growth and crop production 2.
  18. Randelovic, P., V. Dordevic, S. Milic, S. Balesevic-Tubic, K. Petrovic, J. Miladinovic, and V. Dukic, 2020: Prediction of soybean plant density using a machine learning model and vegetation indices extracted from RGB images taken with a UAV. Agronomy 10(8), 1108. https://doi.org/10.3390/agronomy10081108
  19. Redmon, J., S. Divvala, R. Girshick, and A. Farhadi, 2016: You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, 779-788.
  20. RDA (Rural Development Administration of Korea), 2020: A study on the Management and Profitability of Small Area Cultivated Crops
  21. Stehr, N. J., 2015: Drones: The newest technology for precision agriculture. Natural Sciences Education, 44(1), 89-91. https://doi.org/10.4195/nse2015.04.0772
  22. Tack, J., J. Lingenfelser, and S. K. Jagadish, 2017: Disaggregating sorghum yield reductions under warming scenarios exposes narrow genetic diversity in US breeding programs. Proceedings of the National Academy of Sciences 114(35), 9296-9301. https://doi.org/10.1073/pnas.1706383114
  23. Yang, F., and M. Wang, 2021: Deep learning-based method for detection of external air conditioner units from street view images. Remote Sensing 13(18), 3691. https://doi.org/10.3390/rs13183691
  24. Yang, S., L. Hu, H. Wu, H. Ren, H. Qiao, P. Li, and W. Fan, 2021: Integration of crop growth model and random forest for winter wheat yield estimation from UAV hyperspectral imagery. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 14, 6253-6269. https://doi.org/10.1109/JSTARS.2021.3089203
  25. Zhang, Y., M. Li, X. Ma, X. Wu, and Y. Wang, 2022: High-precision wheat head detection model based on one-stage network and GAN model. Frontiers in Plant Science 13.
  26. Zhao, J., X. Zhang, J. Yan, X. Qiu, X. Yao, Y. Tian, Y. Zhu, and W. Cao, 2021: A wheat spike detection method in UAV images based on improved YOLOv5. Remote Sensing 13(16), 3095. https://doi.org/10.3390/rs13163095