Digital Image-Assisted Locomotion Scoring for Dairy Cattle Hooves Using Image Recognition Technology
Taiwan Livestock Research Institute, Ministry of Agriculture
LEE,CHIA-XIN (Assistant Researcher)
Tel：886-037-911696 ext. 232
Email：jxlee@tlri.gov.tw
Taiwan Livestock Research Institute
WANG,SZU-HAN (Associate Researcher)
Apr. 17, 2026
Lameness is one of the three major causes of culling in dairy cattle in Taiwan. According to a study conducted by the Livestock Research Institute, Ministry of Agriculture (2013), the primary reasons for culling dairy cows in Taiwan include milk production disorders (41%), hoof-related diseases (19%), and reproductive disorders (11.6%), indicating that hoof diseases are a critical factor contributing to reduced productive lifespan in dairy cattle. At present, most dairy farms in Taiwan rely on prolonged visual observation and farmers’ experience to identify lame cows. This approach is not only labor-intensive and time-consuming, but also becomes increasingly ineffective as herd sizes have expanded over the past decade. Consequently, cows with hoof problems—particularly those in the early stages of lameness—are difficult to detect, leading to frequent underestimation or neglect of hoof-related disorders.
Lameness reduces the daily feeding time of dairy cattle, as affected animals tend to limit their movement to alleviate pain during walking, resulting in decreased feed intake. These behavioral changes ultimately exert negative effects on milk yield and reproductive performance. Previous studies have demonstrated that early detection and intervention of lameness can significantly improve treatment success rates, prevent progression to chronic hoof diseases, and mitigate the adverse impacts on overall production efficiency.
Locomotion scoring is currently the most widely used method for assessing lameness in dairy cattle. Although scoring systems vary slightly among countries, most are based on changes in gait and posture. In the United Kingdom, the AHDB 0–3 locomotion scoring system is commonly applied, classifying lameness into four levels based on alterations in gait, posture, and weight-bearing patterns. In the United States, the Locomotion Scoring (LS) 1–5 system is widely adopted, categorizing lameness into five levels using similar criteria. Despite their widespread use, traditional manual scoring methods are highly time-consuming, lack objective quantitative standards, and are subject to considerable inter-observer variability due to scorer subjectivity.
With increasing global emphasis on animal welfare, locomotion scoring has become a key indicator for evaluating hoof health in dairy cattle. In addition to Europe and North America, recent efforts in Southeast Asia, Japan, and South Korea have focused on developing automated systems and devices to support long-term health monitoring in dairy herds. Such systems not only address labor shortages on farms but also assist farmers in the early identification of problematic animals, thereby reducing veterinary treatment costs and culling rates. However, commercially available hoof health monitoring systems developed overseas remain expensive, limiting their adoption, particularly among small- and medium-scale dairy farms.
The present study proposes a digital-assisted locomotion scoring approach for dairy cattle based on image analysis and artificial intelligence. Dairy cow gait patterns are initially classified into three categories: normal, mildly lame, and severely lame. Based on the skeletal structure of dairy cattle, 16 identifiable key skeletal landmarks are annotated, with the degree of back arching serving as a primary indicator for model training. In addition, an image database of abnormal hoof-related behaviors is established to support model calibration and accuracy validation during the early stages of system development. This framework enables the construction of a digital image–based locomotion scoring and classification mechanism for automated assessment of hoof health in dairy cattle.
