Desenvolvimento de um aplicativo mobile para o auxílio do teste de tetrazólio nas sementes de soja

• Ano de defesa: 2022
• Autor(a) principal: Mecheln, Luís Henrique Manosso Von
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Medianeira; Brasil; Programa de Pós-Graduação em Tecnologias Computacionais para o Agronegócio; UTFPR
• Palavras-chave em Português: Controle de qualidade; Soja - Biotecnologia; Processamento de imagens; Quality control; Soybean - Biotechnology; Image processing; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Engenharia/Tecnologia/Gestão
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/32447

Soybean cultivation Glycine max (L.) Merr. represents great importance for Brazil, therefore, seed analysis laboratories carry out several tests to measure the quality conditions of the produced lots. Among these tests, the Tetrazolium test stands out, which classifies and evaluates the vigor and viability of planting. However, the visual evaluation stage of the test can generate subjectivity in some cases, in addition to being tiring and tedious for the analysts. The use of computer vision to help identify patterns in seeds has driven research with different image processing techniques and different data classifiers. It is noted that most studies use different sets of images and do not make them available to the scientific community, making it difficult to reproduce the results and compare them with other methods of analysis on the same data. Thus, the purpose of this work is to elaborate a methodology for data collection and development of a database of public images, which will promote scientific research in the ambit of the Tetrazolium test in soybean seeds. The methodology requires a simple approach in the data collection stage, allowing image acquisition and sample classification, altering the final analysis time as little as possible when compared to the traditional method. The agreement of the collected data must be verified between the pairs, where a reassessment is made by other specialists on an image already classified, and then the statistical coefficient of agreement Kappa and ICC of the data is calculated. The implementation of the proposed methodology was carried out with the development of a collection tool and a platform for data availability. As a data collection tool, we used software for mobile devices, such as smartfones and tablets and a WEB system for publication in an image database, offering a verified and segmented base of data and images of soybean seeds that passed the Tetrazolium test.