Weighing live sheep using computer vision techniques and regression machine learning

• DOI: 10.1016/j.mlwa.2021.100076
• Publication Date: 2021-06-19T07:25:04Z
• Authors: Diego André Sant’Ana; Marcio Carneiro B...; José Martins; Wellington Pereir...; Sebastião Lucas N...; Vanir Garcia; Vanessa Aparecida...; Natália da Silva ...; Rodrigo Gonçalves...; Hemerson Pistori
• Keywords: 2. Zero hunger; 0402 animal and dairy science; QA75.5-76.95; 04 agricultural and veterinary sciences; Top-view body area; Image processing; Electronic computers. Computer science; Q300-390; Weight prediction; Body size measurements; Mass estimation; Cybernetics

This research arose from the need to aggregate computer vision technology and machine learning in sheep weight control facilitate weighing process of animals farms. The experiment was conducted collect images their weights, later, annotations were made, generating a mask image dataset. We selected attribute extraction algorithms that extracted shape, size, angles with k-curvature. With these data, we used stratified five-fold cross-validation. Also, eight techniques aimed at regression, result obtained when compared metric Adjusted R2 technique called Random Forest Regressor obtain 0.687 (±0.09) MAE 3.099 (±1.52) kilograms. By performing ANOVA test check if it is statistically relevant using measure, got p-value 0.00000807 (8.07e−06). contribution work prediction non-invasive way images. Therefore, results achieved make possible measure animal's an kg.