Salar Farahmand
CLOVER: Continuous Livestock Observation, Visualization, and Evaluation Resource.
Rel. Roberta Bardini, Alessandro Savino, Paolo Cornale, Lorenzo Martini, Alessia Leclercq. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2026
|
Preview |
PDF (Tesi_di_laurea)
- Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (4MB) | Preview |
Abstract
Daily milk yield is the central indicator of dairy productivity and animal welfare, and unexpected drops often signal disease, heat stress, or nutritional deficits. Reliable short-term forecasts of milk yield can therefore enable timely intervention and prevent economic loss. Precision livestock farming increasingly relies on low-cost behavioral and physiological sensors monitoring activity, rumination, core body temperature, and thermal indices. however, converting these signals into accurate next-day yield forecasts remains an open challenge, particularly for small-to-medium herds that lack the historical yield data automated milking systems normally provide. When historical yield is available, tree-based and linear models normally achieve R²above 0.8-0.9, since yield is strongly autocorrelated day to day.
Using behavioral and physiological sensors alone, however results are considerably weaker with reported R² values ranging from 0.30 to 0.77 depending on herd size and task formulation
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Corso di laurea
Classe di laurea
Aziende collaboratrici
URI
![]() |
Modifica (riservato agli operatori) |
