A modular open-source computer vision pipeline enables individual-level behavior analysis of group-housed pigs
- 1Department of Animal Science, College of Agriculture and Life Sciences, Cornell University, Ithaca, NY, 14853, USA
- 2University of Liège, Gembloux Agro-Bio Tech (ULìege-GxABT), Gembloux, 5030, Belgium
Abstract
Animal behavior analysis is central to understanding welfare, health, and productivity in livestock, yet manual observation is time-consuming, subjective, and difficult to scale. We present a modular pipeline that integrates open-source, state-of-the-art computer vision models to automate individual-level behavior analysis in group-housing environments. The pipeline combines zero-shot object detection, motion-aware segmentation and tracking, and vision-transformer feature extraction for robust behavior recognition, addressing challenges such as occlusion and crowding in indoor pig monitoring. We validated the system on the Edinburgh Pig Behavior Video Dataset across detection, tracking, and behavior-classification tasks. A temporal model achieved 94.2% overall accuracy on nine behaviors, a 21.2 percentage-point improvement over the previous benchmark, while tracking reached 93.3% identity preservation (IDF1) and detection reached 89.3% average precision. The modular design lets individual components be replaced as methods advance and suggests potential for adaptation to other species and settings, although cross-species use would require further validation. By releasing an open-source, end-to-end implementation, this work provides a reproducible and scalable tool for automated, objective, and continuous behavior monitoring in precision livestock farming and welfare assessment.
Related
- A multimodal computer-vision based pipeline for eye-hand coordination quantification during the Nine-Hole Peg Test
- Neuroline: an integrated modular pipeline for FAIR multimodal neurophysiology data processing
- 3D-MAESTRO: A scalable, modular, portable pipeline for automated processing of large-scale volumetric brain microscopy data
- PIGSTI: a modular, reproducible pipeline for detecting species identity, pathogens, and microbes from animal palaeogenomic data
- Reusable modular architecture enables flexible cognitive operations in the mouse brain and artificial recurrent networks
BibTeX
@article{yang2026modular,
title = {{A modular open-source computer vision pipeline enables individual-level behavior analysis of group-housed pigs}},
author = {Haiyu Yang and Enhong Liu and Jennifer Sun and Sumit Sharma and Meike van Leerdam and Sébastien Franceschini and Puchun Niu and Miel Hostens},
journal = {Scientific Reports},
year = {2026},
month = aug,
doi = {10.1038/s41598-026-67803-4},
url = {https://doi.org/10.1038/s41598-026-67803-4},
note = {CC BY},
}
RIS
TY - JOUR
TI - A modular open-source computer vision pipeline enables individual-level behavior analysis of group-housed pigs
AU - Haiyu Yang
AU - Enhong Liu
AU - Jennifer Sun
AU - Sumit Sharma
AU - Meike van Leerdam
AU - Sébastien Franceschini
AU - Puchun Niu
AU - Miel Hostens
PY - 2026
DA - 2026/08/24
JO - Scientific Reports
DO - 10.1038/s41598-026-67803-4
UR - https://doi.org/10.1038/s41598-026-67803-4
AB - Animal behavior analysis is central to understanding welfare, health, and productivity in livestock, yet manual observation is time-consuming, subjective, and difficult to scale. We present a modular pipeline that integrates open-source, state-of-the-art computer vision models to automate individual-level behavior analysis in group-housing environments. The pipeline combines zero-shot object detection, motion-aware segmentation and tracking, and vision-transformer feature extraction for robust behavior recognition, addressing challenges such as occlusion and crowding in indoor pig monitoring. We validated the system on the Edinburgh Pig Behavior Video Dataset across detection, tracking, and behavior-classification tasks. A temporal model achieved 94.2% overall accuracy on nine behaviors, a 21.2 percentage-point improvement over the previous benchmark, while tracking reached 93.3% identity preservation (IDF1) and detection reached 89.3% average precision. The modular design lets individual components be replaced as methods advance and suggests potential for adaptation to other species and settings, although cross-species use would require further validation. By releasing an open-source, end-to-end implementation, this work provides a reproducible and scalable tool for automated, objective, and continuous behavior monitoring in precision livestock farming and welfare assessment.
SN - 2045-2322
ER -