78. Monitoring Dairy Cow Respiration via Lateral Expansion from Dorsal Videos Using Automated Segmentation and Geometric Analysis.

Clicks: 11
ID: 330474
2026
Free PDF
Article Quality & Performance Metrics
Overall Quality
Not rated
Combines reader engagement with the AI quality analysis. This article has not been analysed, so there is no overall score — reader engagement is measured and shown alongside.
AI Quality Assessment
Not analyzed
Readership in this journal
Steady

Ranked #740 of 847 articles by views in italian journal of animal science

Most read Least read

Bar heights use a square-root scale. Only the 120 most-read articles are drawn; the journal has 847 in total.

Mint this article as an NFT
Not yet minted

Create a permanent, verifiable on-chain record of this article on the Scimatic Network. The NFT is held in your Journament account, and you can withdraw it to your own wallet at any time.

5 SUSD one-off · no wallet required
Sign in to mint
Abstract
Abstract Respiration patterns are indicators of health and welfare in dairy cows. However, manual monitoring is labor-intensive and often unfeasible at scale. Existing automated video-based methods for estimating respiration rate are limited to cows in lying positions and rely on one or more deep learning models, which are computationally intensive and require large datasets to ensure generalizability. Therefore, the objective of this study was to develop a training-free pipeline to extract respiration patterns from RGB videos of standing cows using a general-purpose segmentation model and geometric body analysis. During the summer of 2025, we collected one-minute RGB videos from 16 lactating Holstein cows (DIM: 82 ± 21; parity: 2.75 ± 1.3) housed in a tie-stall barn at 8 time points throughout the day (0000, 0230, 0500, 0830, 1200, 1500, 1700, and 2100 h; n = 71 videos), using an Intel RealSense D455 camera at 30 fps from a dorsal view while the cows were standing. To segment the cow’s body, we prompted Segment Anything 2 with a manual polygon on the first frame to track the body across all subsequent frames. We performed principal component analysis on the segmented masks to define the cow’s primary body axes and constructed a transverse width profile from the neck base to the tail. We identified anatomical landmarks to delimit the lumbar area at the widest rib and the hip bones by locating local maxima in the width profile gradient. Finally, we calculated the area under the curve of the width profile between these landmarks over time and reconstructed the respiratory signal using the first 15 Fast Fourier Transform components. The average estimated respiration rate was 40.30 ± 15.07 bpm, which is within reported biological limits (typically 25–100 bpm). We validated the estimates against manual flank movement counts, with r = 0.78 and RMSE = 9.98 bpm. This training-free, geometric approach can be a scalable, robust alternative for automated respiratory monitoring in standing cows, enabling a broad range of applications such as heat stress and health assessment. While the results are promising, the geometric analysis relies on clear anatomical landmarks and may fail when one of the landmarks cannot be located; therefore, future research should compare the performance of this pipeline with a learning-based approach.
Reference Key
openalex_W7214760553 Use this key to autocite in the manuscript while using SciMatic Manuscript Manager or Thesis Manager
Authors Maria Elisa Montes, Guilherme Lobato Menezes, Koops Alexander, Pedro Souto Lamas, Érica Schultz, Caio Alves da Costa, J.r.r. Dórea
Are you one of these authors? Claim your profile
Journal italian journal of animal science
Year 2026
DOI
10.1093/jas/skag272.226
URL
https://doi.org/10.1093/jas/skag272.226
Keywords Keywords not found

Citations

No citations found. To add a citation, contact the admin at info@scimatic.org

Comments

No comments yet. Be the first to comment on this article.