MooSense

BSc (Hons) Artificial Intelligence and Data Science | Data Science Group Project

Menura Gunasiri

Yenuli Sooriyaarachchi

Stephnie Gunasekara

MooSense is an AI-powered smart dairy farm monitoring system that automates herd surveillance and management using computer vision and machine learning. It processes live camera feeds to detect, track and individually identify each cow in real time, removing the need for physical ear tags or constant manual observation. A YOLOv8 five-fold ensemble detects cattle, a centroid-and-IoU tracker maintains their identities across frames, and a biometric Re-Identification module recognises individual cows from their unique appearance, linking every detection to a profile in the herd database. Once identified, each cow’s behaviour — standing, lying, walking, feeding, drinking or licking — is classified and logged over time. These behavioural patterns feed three machine-learning models: a disease-detection model that flags illness early, an estrus-detection model that pinpoints the optimal breeding window and a milk-yield prediction model that forecasts output. Critical events automatically trigger alerts so farmers can intervene promptly. All data is presented through an interactive web dashboard with live surveillance, a biometric herd registry, milk analytics, breeding monitoring, reports, weather integration and remote cloud access for off-site viewing. By combining real-time vision with predictive analytics, MooSense helps dairy farmers improve animal welfare, optimise breeding decisions, increase milk productivity and significantly reduce the labour needed to manage large herds.