Estimation of Tiller Number in Rice Using a Field Robot and Deep Learning

DOI Web Site 21 References
  • SINGH Dhirendranath
    Department of Agricultural and Environmental Engineering, Biotic Environmental Science, The United Graduate School of Agriculture Sciences, Iwate University
  • MORI Tomohiro
    Department of Agricultural and Environmental Engineering, Biotic Environmental Science, The United Graduate School of Agriculture Sciences, Iwate University
  • ICHIURA Shigeru
    Department of Agricultural and Environmental Engineering, Biotic Environmental Science, The United Graduate School of Agriculture Sciences, Iwate University
  • NGUYEN Thanh Tung
    Department of Food, Life and Environment, Faculty of Agriculture, Yamagata University
  • SASAKI Yuka
    Department of Food, Life and Environment, Faculty of Agriculture, Yamagata University
  • KATAHIRA Mitsuhiko
    Department of Food, Life and Environment, Faculty of Agriculture, Yamagata University

Bibliographic Information

Other Title
  • ─Investigating Effects of Dataset Composition on Tiller Estimation Accuracy─

Search this article

Description

Tiller number, an important growth parameter for rice cultivation, is still being assessed manually. This work investigated the influence of dataset composition on performance of deep learning models for tiller number estimation in rice. Four datasets were constructed for early tillering, active tillering, and maximum tillering by applying the concepts of mixed varieties, class balance, and data augmentation. YOLOv4 models were trained to estimate tiller numbers using each constructed dataset. Then their performance was evaluated. Results demonstrated that the models trained with datasets created using a combination of mixed variety, class balance, and augmentation showed the best performance for estimating the tiller number at the three tillering stages with a mAP range of 68.8–86.4 %.

Journal

References(21)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top