Producción de leche real vs la calculada a partir de la ENL estimada por el algoritmo de análisis de imágenes red-green-blue de gramíneas

• Autor(a) principal: Rivera,Oscar Fernando Ospina
• Data de Publicação: 2021
• Outros Autores: Vásquez,Héctor José Anzola, Duarte,Olber Arturo Ayala, Martínez,Andrea Baracaldo, Cantor,Juan Sebastian Arévalo, Arciniegas,Iván Benavides, Perez,Daniel Eduardo Benavides, Enriquez,Gustavo Adolfo Galvis
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Ciência Rural
• Texto Completo: http://old.scielo.br/scielo.php?script=sci_arttext&pid=S0103-84782021000200653
• topic: algorithm; drone; bromatological analysis; RGB; net lactation energy

ABSTRACT: Knowledge about the net lactation energy (NLE) contained in the dry matter of grasses is necessary to make decisions about forage and the balance of diets for grazing cattle. Its determination is made in laboratories using wet or dry chemistry methods, which are costly, delayed, and sometimes present sampling- or process-related reliability problems. An algorithm, which analyzes the red-green-blue (RGB) images of grasses taken by drone, has been developed as a technological alternative. This has allowed us estimating the NLE level, reducing costs, and changing the sampling system and analysis method. The objective of the present study was to compare the milk production, which was calculated from the NLE and estimated using the algorithm for analysis of RGB images of grasses (included in the TaurusWebs® software), vs the actual milk production. The study was conducted in 15 dairy farms belonging to the dairy control system of the Colácteos dairy cooperative, which are located in the upper tropical region (Department of Nariño, Colombia). The prairies evaluated were composed of mixtures of Kikuyo (Pennisetum clandestinum), Raigrás (Lolium spp), and False Poa (Holcus lanatus). The result was analyzed using a linear regression model (R²=0.86; R=0.93). In the Student´s t-test, the actual and estimated milk production averages were equal (P>0.05). In conclusion, the NLE calculated using the algorithm satisfactorily explains the study livestock production, and the information generated by the algorithm can be used to calculate the NLE of grasses.