Exploring the use of an adapted computer vision model to estimate laying hen positions in a cage-free housing system

Authors: AJIBOLA, GIDEON - Purdue University; ORIANS, LOGAN - Purdue University; FERNANDEX-JURICIC, ESTEBAN - Purdue University; KARCHER, DARRIN - Purdue University; FRALEY, GREGORY - Purdue University; ZOLLNER, PATRICK - Purdue University; Jones, Deana; Garcia, Javier; ERASMUS, MARISA - Purdue University

Submitted to: Applied Animal Behaviour Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/18/2026
Publication Date: 4/19/2026
Citation: Ajibola, G., Orians, L., Garcia, J.S., Fernandex-Juricic, E., Jones, D.R., Fraley, G.S., Zollner, P.A., Karcher, D.M., Erasmus, M. 2026. Exploring the use of an adapted computer vision model to estimate laying hen positions in a cage-free housing system. Applied Animal Behaviour Science. https://doi.org/10.1016/j.applanim.2026.107018.
DOI: https://doi.org/10.1016/j.applanim.2026.107018

Interpretive Summary: Cage-free production systems for egg production are becoming more popular in the US egg industry, however it is unknown if these systems meet the behavioral needs of laying hens. Location data of hens can be useful in providing quantifiable metrics for measuring behaviors in the system environment of cage-free hen and how they adapt to this type of production system. Current methods for tracking hens' movement and localizing their positions in cage-free systems involve placing wearable sensors on hens. This method can be costly due to equipment costs. To address these technical and logistical challenges, this study proposed a non-contact computer vision (mapping program) tracking method to estimate hens’ locations in a single-tier cage-free system (n = 200) at 55 and 85 weeks of age. The hens were recorded and the behavioral videos were imported into the mapping program, where a mouse click on the hen generated the hen's spatial coordinates. The x, y coordinates of the hens as they changed their behaviors during 10-min observations and compared the location data to the ground truth (manually estimated coordinates). In this study, the image tracking methodology shows marginal promise for estimating the location of cage-free hens in an indoor production environment thereby requiring further refinement.

Technical Abstract: As the U.S. poultry industry increasingly adopts cage-free housing systems (CFS) for egg production, there is a need to understand how CFS meets the behavioral needs of laying hens, in line with the industry’s welfare commitments. Location data of laying hens can provide quantifiable metrics for measuring the spatial behaviors of cage-free hens and their adaptation to CFS. Contemporary methods of tracking hens and localizing their positions in CFS involve tagging focal hens with wearable sensors. However, current tracking methods for annotating cage-free hen location data are constrained by equipment costs, thereby impacting research methods. To address these technical and logistical challenges, this study proposed a non-contact computer vision (mapping program) tracking method to estimate hens’ locations in a single-tier cage-free system (n = 200) at 55 and 85 WOA. We recorded the hens and imported the behavioral videos into the mapping program, where a mouse click on the focal hen generated the hen's spatial coordinates. We obtained the x, y coordinates of the hens as they changed their behaviors during 10-min observations and compared the location data to the ground truth (manually estimated coordinates). We used a linear mixed effects model (R 12.0) to evaluate the impact of the location estimation methods (manual annotation vs. the mapping program) on laying hen location estimation. The results of our study suggest that the mapping program underestimated the x-axis values of the Cartesian coordinates by a mean difference of 26 cm, while estimation accuracy declined with increasing distance from the camera on the y-axis. Conclusively, the image-based tracking methodology shows marginal promise for estimating the location of cage-free hens in an indoor housing environment, thereby requiring further refinement.