Boosting the performance of SOTA convolution-based networks with dimensionality reduction: An application on hyperspectral images of wine grape berries
Precision viticulture is an area that is very dependent on methods that allow for a sustainable assessment of grape maturity and, in this work, we apply two state-of-the-art (SOTA) convolution-based networks, namely InceptionTime and OmniScale 1D-CNN, to hyperspectral images of wine grape berries to...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-09-01
|
| Series: | Intelligent Systems with Applications |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2667305323000777 |
| _version_ | 1797742641779048448 |
|---|---|
| author | Rui Silva Osvaldo Gramaxo Freitas Pedro Melo-Pinto |
| author_facet | Rui Silva Osvaldo Gramaxo Freitas Pedro Melo-Pinto |
| author_sort | Rui Silva |
| collection | DOAJ |
| description | Precision viticulture is an area that is very dependent on methods that allow for a sustainable assessment of grape maturity and, in this work, we apply two state-of-the-art (SOTA) convolution-based networks, namely InceptionTime and OmniScale 1D-CNN, to hyperspectral images of wine grape berries to estimate sugar content. Since attaining generalization capacity and processing the information in such high-dimensional data are the two biggest challenges to overcome in problems of this nature, we also study the impact of two dimensionality reduction techniques, Principal Component Analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE), on the models' performance. Both models underwent different tests with different vintages and varieties of wine grapes in the training/validation steps, as to form a true test to their generalization capacity. Our results show that both PCA and t-SNE succeed in improving the performance of these deep networks when an adequate number of components is chosen that minimizes the ratio between information loss and removing redundant features: additionally, both techniques significantly reduce computational cost, a very important trait when training deep learning models. Both models showed good generalization ability with very competitive results across different varieties and vintages even despite their significant differences in variability, which is an indicator that a relationship between spectras can be found that is reflected on sugar content values. |
| first_indexed | 2024-03-12T14:43:33Z |
| format | Article |
| id | doaj.art-fa25c970e5ee4c1fa7f0a068a3b55a17 |
| institution | Directory Open Access Journal |
| issn | 2667-3053 |
| language | English |
| last_indexed | 2024-03-12T14:43:33Z |
| publishDate | 2023-09-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Intelligent Systems with Applications |
| spelling | doaj.art-fa25c970e5ee4c1fa7f0a068a3b55a172023-08-16T04:27:29ZengElsevierIntelligent Systems with Applications2667-30532023-09-0119200252Boosting the performance of SOTA convolution-based networks with dimensionality reduction: An application on hyperspectral images of wine grape berriesRui Silva0Osvaldo Gramaxo Freitas1Pedro Melo-Pinto2CITAB - Centre for the Research and Technology of Agro-Environmental and Biological Sciences, Inov4Agro-Institute for Innovation, Capacity Building and Sustainability of Agri-Food Production, Universidade de Trás-os-Montes e Alto Douro, Quinta dos Prados, Vila Real, 5000-801, Portugal; Corresponding author.Centro de Física das Universidades do Minho e do Porto (CF-UM-UP), Universidade do Minho, Rua da Universidade, Braga, 4710-057, Portugal; Departamento de Astronomía y Astrofísica, Universitat de València, Dr. Moliner 50, Burjassot (València), 46100, SpainCITAB - Centre for the Research and Technology of Agro-Environmental and Biological Sciences, Inov4Agro-Institute for Innovation, Capacity Building and Sustainability of Agri-Food Production, Universidade de Trás-os-Montes e Alto Douro, Quinta dos Prados, Vila Real, 5000-801, Portugal; Departamento de Engenharias, Escola de Ciências e Tecnologia, Universidade de Trás-os-Montes e Alto Douro, Quinta dos Prados, Vila Real, 5000-801, PortugalPrecision viticulture is an area that is very dependent on methods that allow for a sustainable assessment of grape maturity and, in this work, we apply two state-of-the-art (SOTA) convolution-based networks, namely InceptionTime and OmniScale 1D-CNN, to hyperspectral images of wine grape berries to estimate sugar content. Since attaining generalization capacity and processing the information in such high-dimensional data are the two biggest challenges to overcome in problems of this nature, we also study the impact of two dimensionality reduction techniques, Principal Component Analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE), on the models' performance. Both models underwent different tests with different vintages and varieties of wine grapes in the training/validation steps, as to form a true test to their generalization capacity. Our results show that both PCA and t-SNE succeed in improving the performance of these deep networks when an adequate number of components is chosen that minimizes the ratio between information loss and removing redundant features: additionally, both techniques significantly reduce computational cost, a very important trait when training deep learning models. Both models showed good generalization ability with very competitive results across different varieties and vintages even despite their significant differences in variability, which is an indicator that a relationship between spectras can be found that is reflected on sugar content values.http://www.sciencedirect.com/science/article/pii/S2667305323000777Hyperspectral imagesWine grape berriesOenological parametersInceptionTimeOmniScale 1D-CNNDimensionality reduction |
| spellingShingle | Rui Silva Osvaldo Gramaxo Freitas Pedro Melo-Pinto Boosting the performance of SOTA convolution-based networks with dimensionality reduction: An application on hyperspectral images of wine grape berries Intelligent Systems with Applications Hyperspectral images Wine grape berries Oenological parameters InceptionTime OmniScale 1D-CNN Dimensionality reduction |
| title | Boosting the performance of SOTA convolution-based networks with dimensionality reduction: An application on hyperspectral images of wine grape berries |
| title_full | Boosting the performance of SOTA convolution-based networks with dimensionality reduction: An application on hyperspectral images of wine grape berries |
| title_fullStr | Boosting the performance of SOTA convolution-based networks with dimensionality reduction: An application on hyperspectral images of wine grape berries |
| title_full_unstemmed | Boosting the performance of SOTA convolution-based networks with dimensionality reduction: An application on hyperspectral images of wine grape berries |
| title_short | Boosting the performance of SOTA convolution-based networks with dimensionality reduction: An application on hyperspectral images of wine grape berries |
| title_sort | boosting the performance of sota convolution based networks with dimensionality reduction an application on hyperspectral images of wine grape berries |
| topic | Hyperspectral images Wine grape berries Oenological parameters InceptionTime OmniScale 1D-CNN Dimensionality reduction |
| url | http://www.sciencedirect.com/science/article/pii/S2667305323000777 |
| work_keys_str_mv | AT ruisilva boostingtheperformanceofsotaconvolutionbasednetworkswithdimensionalityreductionanapplicationonhyperspectralimagesofwinegrapeberries AT osvaldogramaxofreitas boostingtheperformanceofsotaconvolutionbasednetworkswithdimensionalityreductionanapplicationonhyperspectralimagesofwinegrapeberries AT pedromelopinto boostingtheperformanceofsotaconvolutionbasednetworkswithdimensionalityreductionanapplicationonhyperspectralimagesofwinegrapeberries |