Intelligent Classification of Cocoa Cut-Test with Deep Convolutional Neural Networks

• Daniel Benoliel Malcher UEA
• Elloá B. Guedes UEA

DOI: https://doi.org/10.5753/wcama.2022.222804

Abstract

This works aims at addressing the problem of classification of cocoa beans at the Cut-Test as a Computer Vision task with Convolutional Neural Networks. By using a realistic balanced dataset with images labeled by experts into fourteen classes, we carried out an experimental evaluation with holdout cross-validation and repetitions that showed Inception, VGG-16 and EfficientNetB0 as Top-3 architectures with best performance regarding accuracy (68,28 % ± 3,21, 65,54 % ± 3,05 and 55,04 % ± 2,03, respectively). We combined such models into a soft voting ensemble with resulting accuracy of 89,79 % ± 0,92 that outperforms all individual solutions evaluated. This is a promising result that may support strategical goals of improving the quality of cocoa beans production in Brazil.

Keywords: Convolutional Neural Networks, Deep Learning, Digital Agriculture

References

Bahia de Aguiar, P. C. and De Moura Pires, M. (2019). A região cacaueira do sul do estado da Bahia (Brasil): crise e transformação. Cuad. Geogr. Rev. Colomb. Geogr., 28(1):192–208.

Brainer, M. S. D. C. P. (2021). Produção de cacau. Caderno Setorial do Escritório Técnico de Estudos Econômicos do Nordeste, (149):1–23.

Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei, L. (2009). ImageNet: A largescale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition. IEEE.

Dozat, T. (2016). Incorporating Nesterov Momentum into Adam. In Proceedings of the 4th International Conference on Learning Representations, pages 1–4, San Juan, Puerto Rico.

EMBRAPA (2014). Visão 2014–2034 – O Futuro do Desenvolvimento Tecnológico da Agricultura Brasileira (Síntese). Distrito Federal, Brasil, 55p.

Ferreira, A. C. R., Ahnert, D., de Melo Neto, B. A., and Mello, D. L. N. (2013). Guia de Beneficiamento de Cacau de Qualidade. Instituto Cabruca, Ilhéus, Bahia.

Fu, Y. (2020). Image classification via fine-tuning with Efficientnet. Keras. Disponível [link]. Acesso em 11 de maio de 2022.

Haralick, R. M., Shanmugam, K., and Dinstein, I. (1973). Textural features for image classification. IEEE Transactions on Systems, Man, and Cybernetics, SMC-3(6):610–621.

Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., and Adam, H. (2017). MobileNets: Efficient convolutional neural networks for mobile vision applications. Disponível em http://arxiv.org/abs/1704.04861. Acesso em 11 de maio de 2022.

Khan, S., Rahmani, H., Shah, S., and Bennamoun, M. (2018). A Guide to Convolutional Neural Networks for Computer Vision. Number 1 in Synthesis Lectures on Computer Vision. Morgan & Claypool Publishers.

Krizhevsky, A., Sutskever, I., and Hinton, G. E. (2012). ImageNet classification with deep convolutional neural networks. In Proceedings of the 25th International Conference on Neural Information Processing Systems, pages 1097–1105, USA. ACM.

LeCun, Y., , Bottou, L., Bengio, Y., and Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11):2278–2324.

MAPA (2008). Instrução normativa n. 38, de 23 de junho de 2008. Disponível em https://members.wto.org/crnattachments/2008/sps/BRA/08_2038_00_x.pdf. Acesso em 11 de maio de 2022.

MAPA (2022). Comissão executiva do plano da lavoura cacaueira. Disponível em https://www.gov.br/agricultura/pt-br/assuntos/ceplac. Acesso em 11 de maio de 2022.

Polikar, R. (2006). Ensemble based systems in decision making. IEEE Circuits and Systems Magazine, 6(3):21–45.

Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., Berg, A. C., and Fei-Fei, L. (2015). ImageNet Large Scale Visual Recognition Challenge. International Journal of Computer Vision (IJCV), 115(3):211–252.

Saito, S. (2019). Entendendo o relatório de classificação de amêndoas. Centro de Inovação do Cacau. Disponível em [link]. Acesso em 11 de maio de 2022.

Santos, F., Canuto, A., Bedregal, B., Palmeira, E., and Silva, I. (2019a). Supervised methods applied to the construction of a vision system for the classification of cocoa beans in the cut-test. In Anais do XVI Encontro Nacional de Inteligência Artificial e Computacional, pages 72–83, Porto Alegre, RS, Brasil. SBC.

Santos, F., Palmeira, E., and Jesus, G. (2019b). An Image Dataset of Cut-Test-Classified Cocoa Beans. Data in Brief, 24:103916.

Santos, F. A., Palmeira, E. S., and Jesus, G. Q. (2018). Color, structural and textural features for the classification of a Cocoa beans image dataset using Artificial Neural Network. In Anais do XIV Workshop de Visão Computacional, pages 80–84, Ilhéus, Bahia.

Simonyan, K. and Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. Disponível em https://arxiv.org/abs/1409.1556. Acesso em 11 de maio de 2022.

Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., and Rabinovich, A. (2015a). Going deeper with convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition, volume 1, pages 1–9, Boston, Massachusetts, USA. IEEE.

Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., and Wojna, Z. (2015b). Rethinking the inception architecture for computer vision. Disponível em https://arxiv.org/abs/1512.00567. Acesso em 11 de maio de 2022.

Tan, M. and Le, Q. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. In Proceedings of the 36th International Conference on Machine Learning, volume 97, pages 6105–6114.

Tang, S., Zhu, Q., Zhou, X., Liu, S., and Wu, M. (2002). A conception of digital agriculture. In IEEE International Geoscience and Remote Sensing Symposium, pages 3026–3028, Canada. IEEE.

Zhang, C. and Ma, Y. (2012). Ensemble Machine Learning: Methods and Applications. Springer Science+Business Media, USA.