Adrien Kroese

Adrien Kroese

Auto-generated profile, based on publicly available information

Adrien Kroese

Department of Clinical Sciences, Faculty of Veterinary Medicine and Animal Science, Swedish University of Agricultural Sciences, Uppsala, Sweden, 756 51

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 September 2024 in Journal of Dairy Science
Verified
Journal Article
Open access
AKAdrien Kroese
Adrien Kroese
MAMoudud Alam
Moudud Alam
EHElin Hernlund
Elin Hernlund
DBDavid Berthet
David Berthet
LTLena-Mari Tamminen
Lena-Mari Tamminen
NHNiclas Högberg
Niclas Högberg
Freestall comfort is reflected in various indicators, including the ability for dairy cattle to display unhindered posture transition movements in the cubicles. To ensure farm animal welfare, it is instrumental for the farm management to be able to continuously monitor occurrences of abnormal motions. Advances in computer vision have enabled accurate kinematic measurements in several fields, such as human, equine, and bovine biomechanics. An important step upstream to measuring displacement during posture transitions is determining that the behavior is accurately detected. In this study, we propose a framework for detecting lying-to-standing posture transitions from 3-dimensional (3D) pose estimation data. A multiview computer vision system recorded posture transitions between December 2021 and April 2022 in a Swedish stall housing 183 individual cows. The output data consisted of the 3D coordinates of specific anatomical landmarks. The sensitivity of posture transition detection was 88.2%, and precision reached 99.5%. In analyzing those transition movements, breakpoints detected the timestamp of onset of the rising motion, which was compared with that annotated by observers. Agreement between observers, measured by intraclass correlation, was 0.85 between 3 human observers and 0.81 when adding the automated detection. The intra-observer mean absolute difference in annotated timestamps ranged from 0.4 s to 0.7 s. The mean absolute difference between each observer and the automated detection ranged from 1.0 s to 1.3 s. We found a significant difference in annotated timestamps between all observer pairs, but not between the observers and the automated detection, leading to the conclusion that the automated detection does not introduce a distinct bias. We conclude that the model is able to accurately detect the phenomenon of interest and that it is equitable to an observer.

18 July 2025 in Smart Agricultural Technology
Verified
Journal Article
Open access
AKAdrien Kroese
Adrien Kroese
NHNiclas Högberg
Niclas Högberg
EVElena Diaz Vicuna
Elena Diaz Vicuna
DBDavid Berthet
David Berthet
NFNils Fall
Nils Fall
LTLena-Mari Tamminen
Lena-Mari Tamminen

1 July 2026 in Smart Agricultural Technology
Verified
Journal Article
Open access
AKA. Kroese
A. Kroese
YMY. Mellbin
Y. Mellbin
LTL.M. Tamminen
L.M. Tamminen
DBD. Berthet
D. Berthet
ALA. Leclercq
A. Leclercq
MAM. Alam
M. Alam

1 January 2026
Verified
Book
Closed access
AKAdrien Kroese
Adrien Kroese
This thesis sought to apply recent advances in precision livestock farming to evaluating a specific welfare parameter: cows’ comfort when getting up and lying down in cubicles. Cows need sufficient space to get up and lie down but rigid metal bars interfere with their innate motion patterns. A multi-camera system recorded over 12 cubicles during two data-collection phases. Triangulating 24 anatomical landmarks detected on each view using computer vision produced 3D pose estimation throughout posture transitions. From these, the timing, spatial use, and weight distribution could be measured or modelled. When compared with human annotations, the system showed high agreement in identifying rising and lying down events and their phases. While stage detection was not fully repeatable, the results show that 3D keypoint motion reliably reflects observable kinematic patterns. The method developed was then applied to evaluate whether replacing metal head and neck rails with flexible straps improved cows’ movement opportunities. An experiment was run in which the head and neck rails of cubicles were replaced with flexible straps. In the flexible setup, cows showed greater head vertical displacement and straighter lunge during rising, indicating greater movement opportunities. Effect sizes were small. Lying-down movements showed no consistent difference between flexible cubicles and both baselines. The duration of lying down movements increased upon returning to baseline, suggesting that duration alone doesn’t fully capture comfort. There was a consistent difference in a novel indicator introduced in this work: the shift of the cows’ centre of mass. The thesis concludes on the following: (i) Posture transition behaviours consist of multiple, independent dimensions and single-indicator assessments may not be a sound summarisation. (ii) Pose estimation in 3D represents a valuable technology to simultaneously monitor several indicators and uncover different strategies used to cope with restrictive environments. Finally (iii) that novel indicators such as modelled weight displacement are adapted to pose data and get a step closer to biomechanical drivers behind the specific motions and behaviours.