Real-Time Plant Stem Segmentation on Smartphones Using YOLO for Precision Agriculture
Abstract
A crucial aspect of precision agriculture is plant phenotyping, as it provides insights into plant development, status and productivity. In this study, we present an approach for plant stem segmentation optimized for deployment on smartphones to support cost-effective precision agriculture practices. Using the YOLO (You Only Look Once) object detection framework, we trained a custom lightweight model capable of segmenting main stems under variable lighting and plant conditions. The model was evaluated on a dataset collected using a smartphone in a greenhouse, where it achieved 0.727 mAP50 and 20 FPS on Google Pixel 7a smartphone. Therefore, real-time stem segmentation can be effectively integrated into agricultural practices as a practical tool for data-driven crop management.
References
J. Chitwood-Brown, G.E. Vallad, T.G. Lee, S.F. Hutton, Breeding for Resistance to Fusarium Wilt of Tomato: A Review, in Genes, 12 (2021)
Y. Wang, Q. Liu, J. Yang, G. Ren, W. Wang, W. Zhang, F. Li, A Method for Tomato Plant Stem and Leaf Segmentation and Phenotypic Extraction Based on Skeleton Extraction and Supervoxel Clustering, in Agronomy, 14 (2024)
T. Mizik, How can precision farming work on a small scale? A systematic literature review, in Precision Agric, 24 (2023)
Q. Zhang,, R. Luan, M. Wang, J. Zhang, F. Yu, Y. Ping, L. Qiu, Research Progress of Spectral Imaging Techniques in Plant Phenotype Studies, in Plants, 13 (2024)
F. Daichang, X. Lihong, L. Dawei, X. Longjiao, Automatic detection and segmentation of stems of potted tomato plant using Kinect, in Sixth International Conference on Digital Image Processing (ICDIP 2014), 9159 (2014)
F. Zhang, A. Hassanzadeh, J. Kikkert, S.J. Pethybridge, J. v. Aardt, White Mold and Weed Detection in Snap Beans Using UAS-Based Lidar, in IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium, Kuala Lumpur, Malaysia (2022)
L. Loyani, D. Machuve, A Deep Learning-based Mobile Application for Segmenting Tuta Absoluta’s Damage on Tomato Plants, in Eng. Technol. Appl. Sci. Res., 11 (2021)
S. Widiyanto, D.P. Nugroho, A. Daryanto, M. Yunus, D.T. Wardani, Monitoring the Growth of Tomatoes in Real Time with Deep Learning-based Image Segmentation, in International Journal of Advanced Computer Science and Applications(IJACSA), 12 (2021)
L. Zhang, X. Li, Z. Yang, B. Yang, S. Yu, S. Zhao, Z. Huang, X. Zhang, H. Yang, Y. Lin, H. Yu, M. Yang, Tomato Seedling Stem and Leaf Segmentation Method Based on an Improved ResNet Architecture, in Front. Plant Sci., 16 (to be published)
