# An Improved Approach to Automated Measurement of Body Condition Score in Dairy Cows Using a Three-Dimensional Camera System

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## Simple Summary

## Abstract

## 1. Introduction

## 2. Materials and Methods

#### Statistical Analyses

^{2}, does not appear as part of this effective error variance. This was because, under the sampling design of this study, scorer main effects were orthogonal to animal by week and thus not involved in comparisons over weeks nor between animals.

## 3. Results

^{2}× 3 = 41 scorers in this experiment) would have been needed for the visual scoring method to match the performance of the refined camera method in terms of its sensitivity and about five times as many scorers (i.e., 2.1

^{2}× 3 = 14 scorers) to match the performance of the raw camera method. The estimated rate of change, $\beta $, in BCS observed in this experiment is shown in Table 2. It would take 44 days to detect a change in BCS in an animal using visual BCS method but 21 or 12 days to detect a change using the raw or refined camera methods, respectively (Table 2).

## 4. Discussion

^{2})-fold, from 3 to 41 independent visual scorers, to equal the performance of the refined camera. Such was its sensitivity that the refined camera could detect a change in BCS in an estimated 12 days rather than the 44 days for the visual method (Table 2) given the average rate of change observed and the sampling protocols in our experiment. Without excluding outliers, the raw camera method would take an estimated 21 days to detect a change in BCS, and there would need to be RS

^{2}≈ 5 times as many visual scorers to match its sensitivity. Substantial gains in performance were also observed for detecting differences between animals within a week, with RS values of 2.4 for the refined camera method and 1.3 for the raw camera method relative to the visual method. These represent an increased capability of both camera methods for detecting changes and differences in BCS compared with the visual method but especially when raw data were refined as per our statistical method for filtering outliers.

## 5. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## References

- Bewley, J.M.; Schutz, M.M. REVIEW: An Interdisciplinary Review of Body Condition Scoring for Dairy Cattle. Prof. Anim. Sci.
**2008**, 24, 507–529. [Google Scholar] [CrossRef][Green Version] - Roche, J.R.; Friggens, N.C.; Kay, J.K.; Fisher, M.W.; Stafford, K.J.; Berry, D.P. Invited review: Body condition score and its association with dairy cow productivity, health, and welfare. J. Dairy Sci.
**2009**, 92, 5769–5801. [Google Scholar] [CrossRef] [PubMed][Green Version] - Carvalho, P.D.; Souza, A.H.; Amundson, M.C.; Hackbart, K.S.; Fuenzalida, M.J.; Herlihy, M.M.; Ayres, H.; Dresch, A.R.; Vieira, L.M.; Guenther, J.N.; et al. Relationships between fertility and postpartum changes in body condition and body weight in lactating dairy cows. J. Dairy Sci.
**2014**, 97, 3666–3683. [Google Scholar] [CrossRef][Green Version] - Manriquez, D.; Thatcher, W.W.; Santos, J.E.P.; Chebel, R.C.; Galvão, K.N.; Schuenemann, G.M.; Bicalho, R.C.; Gilbert, R.O.; Rodriguez-Zas, S.; Seabury, C.M.; et al. Effect of body condition change and health status during early lactation on performance and survival of Holstein cows. J. Dairy Sci.
**2021**, 104, 12785–12799. [Google Scholar] [CrossRef] [PubMed] - Wildman, E.E.; Jones, G.M.; Wagner, P.E.; Boman, R.L.; Troutt, H.F., Jr.; Lesch, T.N. A dairy cow body condition scoring system and its relationship to selected production characteristics. J. Dairy Sci.
**1982**, 65, 495–501. [Google Scholar] [CrossRef] - Earle, D.F. A guide to scoring dairy cow condition. J. Agric.
**1976**, 74, 228–231. [Google Scholar] - Macdonald, K.A.; Roche, J.R. Condition Scoring Made Easy. Condition Scoring Dairy Herds, 1st ed.; Dexcel Ltd.: Hamilton, New Zealand, 2004; ISBN 0-476-00217-6. [Google Scholar]
- Roche, J.R.; Dillon, P.G.; Stockdale, C.R.; Baumgard, L.H.; VanBaale, M.J. Relationships among international body condition scoring systems. J. Dairy Sci.
**2004**, 87, 3076–3079. [Google Scholar] [CrossRef][Green Version] - Qiao, Y.; Kong, H.; Clark, C.; Lomax, S.; Su, D.; Eiffert, S.; Sukkarieh, S. Intelligent perception for cattle monitoring: A review for cattle identification, body condition score evaluation, and weight estimation. Comput. Electron. Agric.
**2021**, 185, 106143. [Google Scholar] [CrossRef] - Mullins, I.L.; Truman, C.M.; Campler, M.R.; Bewley, J.M.; Costa, J.H.C. Validation of a commercial automated body condition scoring system on a commercial dairy farm. Animals
**2019**, 9, 287. [Google Scholar] [CrossRef] [PubMed][Green Version] - Mandel, J.; Stiehler, R.D. Sensitivity—A criterion for the comparison of methods of test. J. Res. Natl. Bur. Stand.
**1954**, 53, 155–159. [Google Scholar] [CrossRef] - Cleveland, W.S. Robust Locally Weighted Regression and Smoothing Scatterplots. J. Am. Stat. Assoc.
**1979**, 74, 829–836. [Google Scholar] [CrossRef] - R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2020. [Google Scholar]
- Tukey, J.W. Exploratory Data Analysis; Addison-Wesley: Reading, MA, USA, 1977. [Google Scholar]
- Lin, L.I. A Concordance Correlation Coefficient to Evaluate Reproducibility. Biometrics
**1989**, 45, 255–268. [Google Scholar] [CrossRef] [PubMed] - Altman, D.G.; Bland, J.M. Measurement in medicine: The analysis of method comparison studies. Statistician
**1983**, 32, 307–317. [Google Scholar] [CrossRef] - Hannah, M.C.; Giri, K.; Albornoz, R.I.; Gourley, C.J.P.; Awty, I.M.; Wales, W.J. Sensitivity by variance components—A flexible criterion for comparison of measurement methods.
**2021**. in preparation. [Google Scholar] - Galwey, N.W. Introduction to Mixed Modelling, beyond Regression and Analysis of Variance, 2nd ed.; John Wiley & Sons: Chichester, UK, 2014. [Google Scholar]
- Kristensen, E.; Dueholm, L.; Vink, D.; Andersen, J.E.; Jakobsen, E.B.; Illum-Nielsen, S.; Petersen, F.A.; Enevoldsen, C. Within- and across-person uniformity of body condition scoring in Danish Holstein cattle. J. Dairy Sci.
**2006**, 89, 3721–3728. [Google Scholar] [CrossRef] - Francq, B.G.; Govaerts, B. How to regress and predict in a Bland-Altman plot? Review and contribution based on tolerance intervals and correlated-errors-in-variables models. Stat. Med.
**2016**, 35, 2328–2358. [Google Scholar] [CrossRef] [PubMed] - Martins, B.M.; Mendes, A.L.C.; Silva, L.F.; Moreira, T.R.; Costa, J.H.C.; Rotta, P.P.; Chizzotti, M.L.; Marcondes, M.I. Estimating body weight, body condition score, and type traits in dairy cows using three dimensional cameras and manual body measurements. Livest. Sci.
**2020**, 236, 104054. [Google Scholar] [CrossRef]

**Figure 1.**Raw camera body condition score (BCS) values of three selected cows (4617, 3514, and 4612: maximum, median, and minimum BCS mean value cows) by two cameras (open squares and circles) during the experiment, on 1–8 scale. The fitted solid line is a robust loess smooth curve that was used to identify the outliers (red open squares and circles).

**Figure 2.**Visual body condition score (BCS) values of three selected cows (4617, 3514, and 4612: maximum, median, and minimum BCS mean value cows) by three scorers (open squares, open triangles, and inverted open triangles) on 1–8 scale.

**Figure 3.**Scatter plot with each solid dot representing a weekly mean body condition score (BCS) for a cow by refined camera method versus visual measurement method. The solid line represents the line of agreement.

**Figure 4.**Bland–Altman plot with each solid dot representing a weekly mean body condition score (BCS) for a cow between refined camera and visual measurement methods. The central horizontal dotted line represents the mean difference between the two measurement methods, and the fine dotted lines represent the 95% range of the differences. The long-dashed line is a linear regression (p < 0.05) of difference versus the average.

**Table 1.**Estimates of random-effect model parameters (mean and variance components) for each of raw camera, refined camera, and visual body condition score (BCS) data. Camera measurements were taken at least twice daily for a period of 7 weeks on each of 32 cows. The same cows were scored visually for BCS by three scorers on one day of each week over the same period.

Model Parameters | Raw Camera | Refined Camera | Visual Scoring |
---|---|---|---|

Mean BCS | 4.50 | 4.49 | 4.44 |

Variance components (×10^{−2}): | |||

Week (${\sigma}_{W}^{2}$) | 1.70 | 1.90 | 1.00 |

Animal (${\sigma}_{A}^{2}$) | 3.75 | 4.63 | 7.36 |

Camera (${\sigma}_{C}^{2}$), or Scorer $({\sigma}_{S}^{2}$) | 0.00 | 0.00 | 0.04 |

Week.Day $\left({\sigma}_{WD}^{2}\right)$ | 0.05 | 0.04 | |

Week.Animal (${\sigma}_{WA}^{2})$ | 0.52 | 0.77 | 0.94 |

Week.Camera (${\sigma}_{WC}^{2}$), or Week.Scorer (${\sigma}_{WS}^{2}$) | 0.00 | 0.00 | 0.22 |

Animal.Camera (${\sigma}_{AC}^{2}$), or Animal.Scorer (${\sigma}_{AS}^{2}$) | 0.14 | 0.06 | 0.56 |

Week.Day.Milking (${\sigma}_{WDM}^{2})$ | 0.00 | 0.00 | |

Week.Day.Animal (${\sigma}_{WDA}^{2})$ | 0.00 | 0.06 | |

Week.Day.Camera (${\sigma}_{WDC}^{2})$ | 0.00 | 0.00 | |

Week.Animal.Camera (${\sigma}_{WAC}^{2})$ | 0.00 | 0.00 | |

Week.Day.Milking.Animal (${\sigma}_{WDMA}^{2})$ | 0.00 | 0.12 | |

Week.Day.Milking.Camera(${\sigma}_{WDMC}^{2})$ | 0.02 | 0.03 | |

Week.Day.Animal.Camera (${\sigma}_{WDAC}^{2})$ | 0.092 | 0.00 | |

Week.Day.Milking.Animal.Camera (${\sigma}_{WDMAC}^{2})$ | 0.00 | 0.00 | |

Residual (${\sigma}_{\epsilon}^{2})$ | 2.18 | 0.62 | 1.86 |

**Table 2.**Estimates of summary statistics for comparison of raw camera, refined camera, and visual body condition score (BCS) measurement methods. Each statistic relates to the response and/or precision of a cow by week mean. Under both camera methods each mean was an average of approximately 14 camera measurements, and under the visual scoring method, each mean was an average of 3 independent scores.

Summary Statistics | Raw Camera | Refined Camera | Visual Scoring |
---|---|---|---|

For change within animal over time: | |||

Actual SD, ${\sigma}_{a}$, (BCS) | 0.149 | 0.164 | 0.139 |

Error SD, ${\sigma}_{m}$, (BCS) | 0.041 | 0.026 | 0.083 |

Sensitivity, $\Theta $ | 3.6 | 6.2 | 1.7 |

Relative Sensitivity, RS, of camera to visual scoring | 2.1 | 3.7 | |

Rate of change in BCS, β, (BCS/Month) | −0.18 | −0.19 | −0.17 |

Time to detect BCS change of $3{\sigma}_{m}$, (Days) | 21 | 12 | 44 |

For differences between animals at the same week: | |||

Actual SD, ${\sigma}_{a}$, (BCS) | 0.207 | 0.232 | 0.288 |

Error SD, ${\sigma}_{m}$, (BCS) | 0.048 | 0.030 | 0.090 |

Sensitivity $\left(\Theta ={\sigma}_{a}/{\sigma}_{m}\right)$ | 4.3 | 7.8 | 3.2 |

Relative Sensitivity, RS, of camera to visual scoring | 1.3 | 2.4 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Albornoz, R.I.; Giri, K.; Hannah, M.C.; Wales, W.J.
An Improved Approach to Automated Measurement of Body Condition Score in Dairy Cows Using a Three-Dimensional Camera System. *Animals* **2022**, *12*, 72.
https://doi.org/10.3390/ani12010072

**AMA Style**

Albornoz RI, Giri K, Hannah MC, Wales WJ.
An Improved Approach to Automated Measurement of Body Condition Score in Dairy Cows Using a Three-Dimensional Camera System. *Animals*. 2022; 12(1):72.
https://doi.org/10.3390/ani12010072

**Chicago/Turabian Style**

Albornoz, Rodrigo I., Khageswor Giri, Murray C. Hannah, and William J. Wales.
2022. "An Improved Approach to Automated Measurement of Body Condition Score in Dairy Cows Using a Three-Dimensional Camera System" *Animals* 12, no. 1: 72.
https://doi.org/10.3390/ani12010072