HRNet-U: exploiting strengths of U-Net and HRNet for cattle weight prediction
IET Conference Proceedings · 01.08.2026 · 2025 (63) · 216-220
DOI: 10.1049/icp.2026.2364 ↗Kurz gefasst
Fachartikel in der Zeitschrift IET Conference Proceedings, erschienen am 01.08.2026. Verfasst von Mithileshwaran S. und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Accurate estimation of cattle weight plays a vital role in livestock management, health monitoring, and economic valuation. Traditional manual methods for weight measurement are time-consuming, error-prone, and require physical contact with animals, leading to stress and potential injury. This study proposes a deep learning-based, non-invasive approach to predict cattle weight using side and rear-view images. The model integrates a U-Net architecture for precise image segmentation and HRNet for anatomical keypoint detection, including heart girth and body length. The dataset comprises of annotated images that are resized to train and test the U-Net model using standard deep learning techniques. The segmented image along with the heatmap is provided to HRNet to detect 13 keypoints for measurement. The cattle weight is finally predicted from the geometric equation. Experimental results show satisfactory performance, confirming the robustness of the proposed model. The framework offers a reliable alternative to traditional methods and paves the way for practical applications in precision livestock farming.
Zitieren
APA
S., M., P., D., & R., L. R. (2026). HRNet-U: exploiting strengths of U-Net and HRNet for cattle weight prediction. IET Conference Proceedings, 2025(63), 216-220. https://doi.org/10.1049/icp.2026.2364
Vancouver
S. M, P. D, R. LR. HRNet-U: exploiting strengths of U-Net and HRNet for cattle weight prediction. IET Conference Proceedings. 2026;2025(63):216-220. doi:10.1049/icp.2026.2364
BibTeX
@article{S2026,
author = {S., Mithileshwaran and P., Diveshwaran and R., Lekshmi R.},
title = {HRNet-U: exploiting strengths of U-Net and HRNet for cattle weight prediction},
journal = {IET Conference Proceedings},
year = {2026},
volume = {2025},
number = {63},
pages = {216--220},
doi = {10.1049/icp.2026.2364}
}
EUVIMED