Classification of corn productivity using the few-shot learning approach
Estimating productivity is important for agriculture, and machine learning (ML) techniques have contributed to making it happen more quickly and efficiently. Considering the difficulty of acquiring agricultural data on a large scale, few-shot learning (FSL) methods are an alternative. The objective was to evaluate the use of different image composition methods obtained by Remotely Piloted Aircraft, associated or not with plant height, for classifying corn productivity, using traditional and FSL-based AM techniques. The results with FSL showed that the Siamese network model can be viable without using the average plant height.
Classification of corn productivity using the few-shot learning approach
-
DOI: https://doi.org/10.22533/at.ed.317452429017
-
Palavras-chave: -
-
Keywords: -
-
Abstract:
Estimating productivity is important for agriculture, and machine learning (ML) techniques have contributed to making it happen more quickly and efficiently. Considering the difficulty of acquiring agricultural data on a large scale, few-shot learning (FSL) methods are an alternative. The objective was to evaluate the use of different image composition methods obtained by Remotely Piloted Aircraft, associated or not with plant height, for classifying corn productivity, using traditional and FSL-based AM techniques. The results with FSL showed that the Siamese network model can be viable without using the average plant height.
- Gabriel Tonon Cimatti
- Alaine Margarete Guimarães
- Eduardo Fávero Caires
- Gabriel Passos de Jesus
