ECO4Horse - An AI-based Platform for Analyzing Horse Health using video and sensor data
Ingénierie et Architecture
In equestrian sports and horse training, ensuring animal welfare and peak performance depends on the early detection of health issues. Traditional monitoring methods rely heavily on human observation and periodic veterinary assessments, which can overlook subtle or early signs of distress. This limitation creates a need for more objective, continuous, and automated approaches to health monitoring.
Recent advances in wearable sensors, such as accelerometers, and video capture technologies, combined with the power of Artificial Intelligence (AI), offer a unique opportunity to transform horse health management. By leveraging these tools, this project bridges the gap between raw data collection and actionable medical insights, ultimately improving care and performance.
Our long-term vision is to develop AI algorithms capable of automatically detecting and classifying health conditions in horses by integrating time-series data from sensors with video analysis. While current solutions, such as Alogo Move Pro, provide anomaly detection through time-series analysis, they fall short of delivering comprehensive medical diagnostics.
Achieving this goal requires overcoming several challenges:
- Data Integration and Synchronization: Building a platform that retrieves both accelerometry and video data, synchronizes them accurately, and stores them on an edge server for further processing.
- Expert-Labeled Dataset Creation: Establishing a robust data acquisition process where expert riders and veterinarians label synchronized data to create a high-quality dataset for AI development.
- AI Algorithm Development: Designing advanced models capable of diagnosing horse health conditions in real-world training and competition environments.
The ECO4Horse project addresses this challenges by developing an edge-to-cloud platform that enables real-time collection, synchronization, and pre-processing of accelerometry data from Alogo Move Pro and high-resolution video captured via smartphones. Video data undergoes a machine learning-based preprocessing (cropping, refocusing, resizing, and automatic selection of regions of interest) and pose estimation techniques are applied for segmentation and classification, to later-on enable anomaly detection.
To facilitate expert labeling, the project creates a smartphone application allowing veterinarians and horse specialists to efficiently annotate data, ensuring accuracy while minimizing time requirements. The resulting labeled dataset serves as the foundation for future AI-driven diagnostic systems.
By combining cutting-edge technology with expert knowledge, ECO4Horse aims to revolutionize equine health monitoring, paving the way for intelligent diagnostic solutions that enhance animal welfare and optimize performance in competitive and training contexts.
Un appel à projets destiné aux entreprises pour stimuler leur innovation, leur compétitivité et leur durabilité
Le projet décrit ci-dessus fait partie d'un appel à projets extraordinaire intitulé "Innovation, compétitivité et durabilité", lancé par le domaine Ingénierie et Architecture de la HES-SO. Ces fonds sont destinés aux professeur-es proches des sociétés de services et d'entreprises suisses. Les hautes écoles concernées par cet appel à projets sont :
- HE-Arc Ingénierie ;
- Haute école d'ingénierie et d'architecture de Fribourg - HEIA-FR ;
- Haute école du paysage, d'ingénierie et d'architecture de Genève (HEPIA) ;
- HES-SO Valais-Wallis - Haute Ecole d'Ingénierie - HEI ;
- Haute Ecole d'Ingénierie et de Gestion du Canton de Vaud - HEIG-VD ;
- CHANGINS - Haute école de viticulture et œnologie.