Published:
2025-06-25Issue:
Vol. 19 No. 1 (2025)Section:
Applied Engineering VisionSmart System for Ripeness Detection in Blackberry Fruits
Sistema Inteligente de Detección de Madurez en Frutos de Mora
Keywords:
Convolutional Neural Network, TensorFlow Lite, Blackberry Classification, Embedded Systems, Deep Learning, Ripeness Detection (en).Keywords:
Red Neuronal Convolucional, Tensorflow Lite, Clasificación de Mora, Sistemas Embebidos, Aprendizaje Profundo, Detección de Madurez (es).Downloads
Abstract (en)
This paper presents the development of an algorithm aimed at enhancing the classification process of blackberry fruits, which are traditionally harvested manually based on the experience of the workers. The algorithm leverages a VGG-16 convolutional neural network to recognize the ripeness level of blackberry fruits. To achieve this, a comprehensive image dataset was created to train the neural network. The network was designed, trained, and optimized based on specific performance parameters and then implemented on a Raspberry Pi using the TensorFlow Lite framework. Field tests were conducted on actual blackberry crops to evaluate the algorithm's performance, showing a high accuracy rate of 96.66%. The results highlight the potential of the proposed system to significantly improve the efficiency and accuracy of the fruit classification process, reducing the reliance on manual methods and enabling more consistent harvesting practices.
Abstract (es)
Este trabajo presenta el desarrollo de un algoritmo destinado a mejorar el proceso de clasificación de los frutos de mora, que tradicionalmente se recolectan manualmente basándose en la experiencia de los trabajadores. El algoritmo aprovecha una red neuronal convolucional VGG-16 para reconocer el nivel de madurez de los frutos de mora. Para ello, se creó un amplio conjunto de datos de imágenes para entrenar la red neuronal. La red se diseñó, entrenó y optimizó en función de parámetros de rendimiento específicos y, a continuación, se implementó en una Raspberry Pi utilizando el marco TensorFlow Lite®. Se realizaron pruebas de campo en cultivos reales de mora para evaluar el rendimiento del algoritmo, que mostró una elevada tasa de precisión del 96,66 %. Los resultados destacan el potencial del sistema propuesto para mejorar significativamente la eficiencia y la precisión del proceso de clasificación de frutas, reduciendo la dependencia de los métodos manuales y permitiendo prácticas de cosecha más consistentes.
References
N. Aghilinategh, M. J. Dalvand, and A. Anvar, “Detection of ripeness grades of berries using an electronic nose,” Food Sci. Nutr., vol. 8, no. 9, pp. 4919–4928, Sep. 2020, doi: 10.1002/fsn3.1788.
M. Rizzo, M. Marcuzzo, A. Zangari, A. Gasparetto, and A. Albarelli, “Fruit ripeness classification: A survey,” Artif. Intell. Agric., vol. 7, pp. 44–57, 2023, doi: 10.1016/j.aiia.2023.02.004.
V. Maharshi, S. Sharma, R. Prajesh, S. Das, A. Agarwal, and B. Mitra, “A Novel Sensor for Fruit Ripeness Estimation Using Lithography Free Approach,” IEEE Sens. J., vol. 22, no. 22, pp. 22192–22199, Nov. 2022, doi: 10.1109/JSEN.2022.3210439.
S. Tian and H. Xu, “Mechanical-based and Optical-based Methods for Nondestructive Evaluation of Fruit Firmness,” Food Rev. Int., vol. 39, no. 7, pp. 4009–4039, Aug. 2023, doi: 10.1080/87559129.2021.2015376.
A. Lazaro, M. Boada, R. Villarino, and D. Girbau, “Color Measurement and Analysis of Fruit with a Battery-Less NFC Sensor,” Sensors, vol. 19, no. 7, p. 1741, Apr. 2019, doi: 10.3390/s19071741.
D. Wang, M. Zhang, A. S. Mujumdar, and D. Yu, “Advanced Detection Techniques Using Artificial Intelligence in Processing of Berries,” Food Eng. Rev., vol. 14, no. 1, pp. 176–199, Mar. 2022, doi: 10.1007/s12393-021-09298-5.
ICONTEC, Frutas frescas. Mora de castilla. Especificaciones - NTC 4106. Colombia: Instituto Colombiano de Normas Técnicas y Certificación, 1997.
S. Tewari, “CNN Architecture Series — VGG-16 with implementation (Part I),” Medium, 2019. https://medium.com/datadriveninvestor/cnn-architecture-series-vgg-16-with-implementation-part-i-bca79e7db415.
H. M. Rai and K. Chatterjee, “Detection of brain abnormality by a novel Lu-Net deep neural CNN model from MR images,” Mach. Learn. with Appl., vol. 2, p. 100004, Dec. 2020, doi: 10.1016/j.mlwa.2020.100004.
How to Cite
APA
ACM
ACS
ABNT
Chicago
Harvard
IEEE
MLA
Turabian
Vancouver
Download Citation
Visitas
Downloads
License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
1.png)
Attribution-NonCommercial 4.0 International
.jpg)
