Helena Russello

Wageningen University &
Research Farm Technology Group

EDL P16-25 P5: Deep Learning for Human and Animal Health (HAAH)

Research assignment
Automated gait analysis of dairy cows

Dairy farmers perform various tasks around the farm and, with increasingly large herds, are left with very little time to watch their cows. As a result, health issues such as hoof lesions are often detected too late. If left untreated, hoof lesions can lead to severe complications, resulting in lower milk production and a larger chance of early culling. Dairy farms could highly benefit from automated cow monitoring.
This project focuses on researching computer vision methods for lameness detection, that is the detection of abnormal gait patterns that could indicate painful hoof lesions.
Using videos of walking cows, their body parts are localized in each frame using deep learning pose estimation models. Then, the spatio-temporal movement of the body parts is analyzed to analyze the cows’ gait.
By using gait monitoring, abnormal gaits can be detected early and allow farmers to treat lesions in time, thus maintaining production levels while improving animal welfare.


T-LEAP: Occlusion-robust pose estimation of walking cows using temporal information. Helena Russello, Rik van der Tol, Gert Kootstra. Journal publication in: Computers and Electronics in Agriculture.
Public access-gold.

Personal information