Evaluating Alternatives to Locomotion Scoring for Detecting Lameness in Pasture-Based Dairy Cattle in New Zealand: In-Parlour Scoring

Development of a real-time cattle lameness detection system using a single side-view camera

Recent advancements in machine learning and deep learning have revolutionized various computer vision applications, including object detection, tracking, and classification. This research investigates the application of deep learning for cattle lameness detection in dairy farming. Our study employs image processing techniques and deep learning methods for cattle detection, tracking, and lameness classification. We utilize two powerful object detection algorithms: Mask-RCNN from Detectron2 and the popular YOLOv8. Their performance is compared to identify the most effective approach for this application. Bounding boxes are drawn around detected cattle to assign unique local IDs, enabling individual tracking and isolation throughout the video sequence. Additionally, mask regions generated by the chosen detection algorithm provide valuable data for feature extraction, which is crucial for subsequent lameness classification. The extracted cattle mask region values serve as the basis for feature extraction, capturing relevant information indicative of lameness. These features, combined with the local IDs assigned during tracking, are used to compute a lameness score for each cattle. We explore the efficacy of various established machine learning algorithms, such as Support Vector Machines (SVM), AdaBoost and so on, in analyzing the extracted lameness features. Evaluation of the proposed system was conducted across three key domains: detection, tracking, and lameness classification. Notably, the detection module employing Detectron2 achieved an impressive accuracy of 98.98%. Similarly, the tracking module attained a high accuracy of 99.50%. In lameness classification, AdaBoost emerged as the most effective algorithm, yielding the highest overall average accuracy (77.9%). Other established machine learning algorithms, including Decision Trees (DT), Support Vector Machines (SVM), and Random Forests, also demonstrated promising performance (DT: 75.32%, SVM: 75.20%, Random Forest: 74.9%). The presented approach demonstrates the successful implementation for cattle lameness detection. The proposed system has the potential to revolutionize dairy farm management by enabling early lameness detection and facilitating effective monitoring of cattle health. Our findings contribute valuable insights into the application of advanced computer vision methods for livestock health management.

Hind feet position score: A novel trait to genetically reduce lameness incidence

Lameness is an important health and welfare issue that causes considerable economic losses in dairy herds. The objective of this study was to investigate whether the hind feet position score (HFPS) can be used as an auxiliary trait for genetic evaluation of lameness. The HFPS is evaluated by visual scoring of the position of both the hind-digits to the mid-line of the cow's body. The higher the heel height of the lateral claw, the higher is the HFPS, and the higher is the risk for development of lameness. In total, 3,478 records from 1,064 Fleckvieh cows from 35 farms were obtained between September 1, 2021, and March 5, 2022. Data collection was carried out by the regional milk recording organizations. Hind feet position was scored visually by trained personnel during routine milk performance testing in the milking parlor using a 3-class scoring system: score 1 = 0° to <17° indicating a balanced heel height of both the medial and the lateral claw; score 2 = angle of 17° to 24°; score 3 = angle of >24°. After all cows had been milked, locomotion scoring was performed for each animal using a 5-class scoring system with locomotion scores ranging between 1 (normal) and 5 (severely lame). Using HFPS, sensitivity and specificity were 69.5% and 66.8%, respectively, for detecting lameness defined by locomotion score ≥2. For genetic analyses, a bivariate linear animal model was fitted with fixed effects of herd, parity, lactation stage, and classifier, and random effects of animal and permanent environment. Heritabilities for HFPS and locomotion score were 0.07 and 0.10, respectively, and the genetic correlation between the 2 traits studied was 0.80. These results suggest that the HFPS could be used for genetic evaluations to reduce lameness incidence in dairy cattle.

Investigating the effect of prophylactic claw trimming on the interval between calving and first observed elevated locomotion score in pasture-based dairy cows

AIMS

To evaluate, in a pasture-based dairy herd, the response to a three-time point hoof trimming regime on lameness incidence and time from calving to observation of an elevated locomotion score (LS).

METHODS

This study was conducted on a 940-cow spring-calving herd in New Zealand's North Island between May 2018 and May 2019. Cows (n = 250) were randomly allocated to the hoof trimming group, with the remainder assigned to the non-trim cohort. One trained professional hoof trimmer used the five-step Dutch method to trim the hind feet of the trimming group. Throughout the subsequent production season, the whole herd was locomotion-scored fortnightly using the 4-point (0-3) Dairy NZ lameness score. Kaplan-Meier survival curves were used to assess the univariable effect of trimming on the interval between calving and first LS of ≥ 2 and first LS ≥ 1. A multivariable Cox proportional hazards regression was used to further evaluate the effect of trimming on time to elevated LS.

RESULTS

Mean lameness (LS ≥ 2) prevalence was 2.6%, with 30% of cows having ≥ 4 observations during the study period when at least one LS was ≥ 2. For LS ≥ 1, mean prevalence was 40%, with 98.6% of cows having ≥ 4 observations during the study period when at least one LS was ≥ 1 during lactation. Hoof trimming had no apparent effect on the incidence of clinical lameness (LS ≥ 2) (trimmed vs. non-trimmed: 33.2% vs. 28.8%, respectively), but for LS ≥ 1, there was a small decrease in the incidence of LS ≥ 1 (trimmed vs. non-trimmed: 96.9% vs. 99.3%, respectively). The hazard of a cow having a first observed LS ≥ 2 in the control group was 0.87 (95% CI = 0.66-1.14) times that of the trimmed group; however, the hazard of a cow having a first LS ≥ 1 was 1.60 (95% CI = 1.37-1.88) times higher in the control than in the trimmed group.

CONCLUSION AND CLINICAL RELEVANCE

On this farm, prophylactic hoof trimming had no clinically relevant impact on the incidence of clinical lameness and was not associated with clinically beneficial reductions in time to first observed LS ≥ 2. This may be because claw horn imbalance was not pronounced on this farm, with 53% of cows needing no trim on either hind limb on the first trimming occasion. Further research on the response to prophylactic trimming in pasture-based dairy cattle is required.

Review: Assessment of dairy cow welfare at pasture: measures available, gaps to address, and pathways to development of ad-hoc protocols

Pasture is generally perceived as positive for dairy cow welfare, but it nevertheless exposes cows to heat, parasites, and other challenges. This review is intended for people ready to design comprehensive protocols for assessing the welfare of dairy cows at pasture. We provide an overview of the benefits and risks of pasture for cows, and then go on to identify the available and feasible measures for assessing cow welfare at pasture and the gaps that need to be addressed to develop specific welfare measures. Some of the measures from on-farm welfare assessment protocols designed for indoor use (e.g. Welfare Quality®) are relevant for cows at pasture (e.g. lameness scoring). However, the timing, location and/or method for certain measures (e.g. observation of social behaviour) need to be adapted to the pasture context, as cows at pasture can roam over a large area. Measures to address specific pasture-related risks (e.g. heat stress, biosecurity) or benefits (e.g. expression of a wide range of behaviours) should be implemented in order to capture all dimensions of cow welfare at pasture. Furthermore, cow welfare is liable to vary over the grazing season due to changes in weather conditions, grass quality and pasture plots that induce variations in lying surface conditions, food availability, distance to walk to the milking parlour, and so on. It is therefore important to investigate the variability in different welfare measures across the pasture season to check whether they hold stable over time and, if not, to determine solutions that can give an overview across the grazing season. Sensors offer a promising complement to animal and environment observations, as they can capture long-term animal monitoring data, which is simply not possible for a one-day welfare-check visit. We conclude that some measures validated for indoor situations can already be used in pasture-based systems, while others need to be validated for their fitness for purpose and/or use in pasture conditions. Furthermore, thresholds should probably be determined for measures to fit with pasture contexts. If all measures can be made adaptable to all situations encountered on farms or variants of the measures can at least be proposed for each criterion, then it should be possible to produce a comprehensive welfare assessment protocol suitable for large-scale use in near future.

Performance of Various Filtering Media for the Treatment of Cow Manure from Exercise Pens—A Laboratory Study

During summer and winter months, pastures and outdoor pens represent the conventional means of providing exercise for dairy cows housed in tie-stall barns in the province of Québec, Canada. Unfortunately, outdoor pens require large spaces, and their leachates do not meet Québec’s environmental regulations. Therefore, there is a need to develop alternative approaches for these so-called wintering pens. A sustainable year-long approach could be a stand-off pad consisting of a filtering media to manage adequately water exiting the pad. Different filtering materials can be used and mixed (gravel, woodchips, biochar, sphagnum peat moss, sand, etc.). To find the best material and/or mixes, a laboratory study was carried out using 15 PVC pipes (5 cm in diameter and 50 cm long) to test five different combinations of materials over a 3-week period. Different contaminant-removal efficiencies were achieved with the alternative materials, including for chemical oxygen demand (11–38%), phosphates (8–23%), suspended solids (33–57%), and turbidity (23–58%). Alternative treatments with sand, sphagnum peat moss, and biochar improved the filtration capacity when compared to the conventional material (woodchips). However, after three weeks of experimentation, the treatment efficiency of sand gradually decreased for pollutants such as suspended solids and phosphates.