Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 3
Encontrada(s) 1 página(s)

DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Novel approach based on artificial intelligence to evaluate individual wine intake, Un nuevo método basado en inteligencia artificial para evaluar la ingesta individual de vino

  • Cobo Cano, Miriam
  • Relaño de la Guía, Edgard
  • Heredia, Ignacio
  • Aguilar, Fernando
  • Lloret Iglesias, Lara
  • García Díaz, Daniel
  • Yuste, Silvia
  • Recio-Fernández, Emma
  • Pérez-Matute, Patricia
  • Motilva, María-José
  • Moreno-Arribas, M. Victoria
  • Bartolomé, Begoña
Trabajo presentado en el 44th World Congress Vine & Wine, celebrado en Cádiz-Jerez (España), de 5 al 9 de junio de 2023, [EN] This study arises from the need to propose new methodologies to quantify wine consumption more precisely in order to use subsequently this information in observational food-health studies and dietary intervention studies. It has been developed an algorithm based on a “deep learning” method to determine wine volume from a single-view image, and it has been validated through a consumer study developed via a web application. The new model demonstrated satisfactory performance not only in a “daily lifelike” images dataset but also in “real” images (obtained from the consumer study), with a mean absolute error (MAE) of 10 and 26 mL, respectively. In relation to the data reported by the participants in the consumer study (n=38), average red wine volume in a glass was 114±33 mL, without being affected by factors such as gender, time of consumption, type of wine or type of glass. Therefore, the deep learning system together with the web application developed in this study constitute a diet monitoring tool of substantial value in the accurate assessment of daily wine intake, as well as in the habits of its consumption, with relevant applications in observational studies., [ES] Este estudio surge de la necesidad de nuevas metodologías que permitan cuantificar el consumo de vino con mayor precisión, para posteriormente utilizar esta información en estudios observacionales de alimentación-salud y estudios de intervención de dieta. Se ha desarrollado un algoritmo basado en un método de “aprendizaje profundo”, que permite determinar el volumen de vino en una copa/vaso a partir de una fotografía, y se ha validado en un estudio de consumidores realizado a través de una aplicación web. La aplicación del modelo a imágenes “cuasi-reales” y a imágenes "reales" (obtenidas a partir del estudio de consumidores), ha mostrado una precisión satisfactoria con un error absoluto medio (MAE) de 10 mL y 26 mL, respectivamente. En relación a las pautas de consumo de vino observadas en el estudio de consumidores (n=38), el volumen medio de vino tinto servido en una copa fue de 114±33 mL, sin estar condicionado por factores como el sexo del consumidor, el momento de consumo, el tipo de vino, o el formato de copa/vaso. En síntesis, el sistema de aprendizaje profundo desarrollado junto con la aplicación web, constituyen una herramienta de gran valor para la estimación precisa del volumen de vino consumido diariamente, así como las pautas de su consumo, de gran utilidad para estudios poblacionales, Este trabajo ha sido financiado por el Ministerio de Ciencia e Innovación (MCIN) a través de la Agencia Estatal de Investigación (AEI)/10.13039/501100011033 y por la “Unión Europea Generation EU/PRTR”: proyectos PID2019- 108851RB-C21 y PID2019-108851RB-C22, y 'Prueba de Concepto' PDC2022-133861-C21 y PDC2022-133861-C22. M. Cobo agradece el apoyo del Consejo Superior de Investigaciones Científicas (CS, Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Novel digital-based approach for evaluating wine components’ intake: A deep learning model to determine red wine volume in a glass from single-view images

  • Cobo Cano, Miriam
  • Relaño de la Guía, Edgard
  • Heredia, Ignacio
  • Aguilar, Fernando
  • Lloret Iglesias, Lara
  • Yuste, Silvia
  • Recio-Fernández, Emma
  • Pérez-Matute, Patricia
  • Motilva, María-José
  • Moreno-Arribas, M. Victoria
  • Bartolomé, Begoña
  • García Díaz, Daniel
Estimation of wine components’ intake (polyphenols, alcohol, etc.) through Food Frequency Questionnaires (FFQs) may be particularly inaccurate. This paper reports the development of a deep learning (DL) method to determine red wine volume from single-view images, along with its application in a consumer study developed via a web service. The DL model demonstrated satisfactory performance not only in a daily lifelike images dataset (mean absolute error = 10 mL), but also in a real images dataset that was generated through the consumer study (mean absolute error = 26 mL). Based on the data reported by the participants in the consumer study (n = 38), average red wine volume in a glass was 114 ± 33 mL, which represents an intake of 137–342 mg of total polyphenols, 11.2 g of alcohol, 0.342 g of sugars, among other components. Therefore, the proposed method constitutes a diet-monitoring tool of substantial utility in the accurate assessment of wine components’ intake., This study was supported by MCIN (Ministerio de Ciencia e Innovación)/AEI (Agencia Estatal de Investigación)/ 10.13039/501100011033 and the European Union NextGenerationEU/PRTR through the projects PID2019-108851RB-C21 and PID2019-108851RB-C22, and ‘Prueba de Concepto’ PDC2022-133861-C21 and PDC2022-133861-C22. M. Cobo acknowledges support from Consejo Superior de Investigaciones Científicas (CSIC) and Institute of Physics of Cantabria (IFCA) through the CSIC Interdisciplinary platform (PTI+) Global Health, and the Ministry of Education of Spain (FPU grant, reference FPU21-04458 ). I. Heredia was supported by the Gobierno de Cantabria with the project “Instrumentación y Ciencia de Datos para sondear la naturaleza del Universo”. The authors would like to thank CSIC Interdisciplinary Thematic Platform (PTI+) Digital Science and Innovation)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Fruit weight estimation using image-based deep learning for use in dietary assessment

  • Izquierdo González, Pablo
  • Aragón Espinosa, Patricia
  • Relaño de la Guía, Edgard
  • Cobo Cano, Miriam
  • Heredia, Ignacio
  • García Díaz, Daniel
  • Aguilar, Fernando
  • Lloret Iglesias, Lara
  • Yuste, Silvia
  • Íñiguez, María
  • Pérez-Matute, Patricia
  • Moreno-Arribas, M. Victoria
  • Bartolomé, Begoña
  • Motilva, María-José
One of the key limitations of current dietary assessment methodologies is to adjust the food portion size. In this study, we propose a Deep Learning (DL) approach to estimate the weight of individual pieces of commonly consumed fruits (apple, pear, orange and banana) from single-view RGB (Red, Green and Blue) photographs. The DL models developed in this study were based on convolutional neural networks trained on an ad-hoc dataset of 48,960 photographs including a wide and representative range of fruit piece weights, reflecting typical market variability and reducing the likelihood of encountering out-of-range samples. The photographs of apples (n=12,960), pears (n=17,208), oranges (n=8,712) and bananas (n=10,080) were taken under different conditions. The DL models were evaluated in terms of saliency maps and regression metrics. Mean Absolute Error (MAE) values indicated that the measurements in the DL models developed would be out —as a mean— by no more than 20.57 g (apple), 19.25 g (pear), 28.31 g (orange) and 21.93 g (bananas) between the predicted and the observed values, quite acceptable considering the variability in fruit weights and photographic conditions. The four DL models developed in this study predict fruit weight from a simple photograph, requiring only a visible €1 coin as a reference, without considering the background. This approach is feasible because our DL models were trained on diverse images across different angles, distances, lighting conditions, tablecloths and dish types, allowing the models to generalise well in real-world contexts., This study was supported by the MCIN (Ministerio Español de Ciencia e Innovación)/AEI (Agencia Estatal de Investigación)/10.13039/501100011033 and the European Union NextGenerationEU/PRTR through the projects ‘Prueba de Concepto’ PDC2022–133861-C21 and PDC2022–133861-C22., Peer reviewed