Evaluating multiple regressors for the yield of orange orchards

• Kleber X. S. de Souza Embrapa
• Sônia Ternes Embrapa
• João Camargo Neto Embrapa
• Thiago T. Santos Embrapa
• Alécio Souza Moreira Embrapa
• Luciano V. Koenigkan Embrapa
• Roberta de Souza Embrapa

DOI: https://doi.org/10.5753/sbiagro.2023.26567

Resumo

Accurate fruit yield estimation is crucial for making informed decisions about harvesting, storage and marketing. However, estimating fruit yield can be challenging. Currently, yield estimation relies on labor-intensive manual counting combined with statistical methods. However, computer vision has emerged as a potential alternative by enabling automatic fruit counting, thus simplifying the process. In this paper, we assess the effectiveness of various machine learning regressors for yield forecasting based on fruit detection in images captured within the orchard. Our results indicate that deep forward neural networks outperform the other regressors examined, making them the most effective choice for yield prediction.

Referências

Breiman, L. (2001). Random forests. Machine Learning, 45(1):5–32.

Camargo Neto, J., Ternes, S., de Souza, K. X. S., Yano, I. H., and Queiros, L. R. (2019). Uso de redes neurais convolucionais para detecção de laranjas verdes. In ANAIS DO CONGRESSO BRASILEIRO DE AGROINFORMÁTICA, Indaiatuba, SP.

Cerqueira, L. M., de Souza, K. X. S., Ternes, S., and Camargo Neto, J. (2020). Usando a rede neural faster-rcnn para identificar frutos verdes em pomares de laranja. In CONGRESSO INTERINSTITUCIONAL DE INICIAÇÃO CIENTÍFICA, Campinas, SP. Embrapa Informática Agropecuária.

Cortes, C. and Vapnik, V. (1995). Support-vector networks. Machine learning, 20(3):273–297.

Drucker, H., Chris, Kaufman, B. L., Smola, A., and Vapnik, V. (1997). Support vector regression machines. In Advances in Neural Information Processing Systems 9, volume 9, pages 155–161.

Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. The Annals of Statistics, 29(5).

Fundecitrus (2022). Relatório de atividades: junho 2021/maio 2022. Technical report, Fundo de Defesa da Citricultura (Fund for Citrus Protection) – Fundecitrus.

Géron, A. (2019). Hands-on Machine Learning with Scikit-Learn, Keras, and Tensor-Flow: Concepts, Tools, and Techniques to Build Intelligent Systems. O’Reilly Media, Incorporated.

Goodfellow, I., Bengio, Y., and Courville, A. (2017). Deep Learning. MIT Press.

Häni, N., Roy, P., and Isler, V. (2020). A comparative study of fruit detection and counting methods for yield mapping in apple orchards. Journal of Field Robotics, 37(2):263–282.

King, A. P. and Zisserman, A. (2019). Statistics for Biomedical Engineers and Scientists. Academic Press, first edition.

Kohavi, R. (1995). A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection. In Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI’95), pages 1137–1143.

Koirala, A., Walsh, K., Wang, Z., and McCarthy, C. (2019). Deep learning for real-time fruit detection and orchard fruit load estimation: Benchmarking of ‘MangoYOLO’. Precision Agriculture, 20:1107–1135.

Maldonado, W. and Barbosa, J. C. (2016). Automatic green fruit counting in orange trees using digital images. Computers and Electronics in Agriculture, 127:572–581.

Sousa, M. A., de Souza, K. X. S., Camargo Neto, J., Ternes, S., and Yano, I. H. (2021). Usando a rede neural ssd para identificar frutos verdes em pomares de laranja. In CONGRESSO INTERINSTITUCIONAL DE INICIAÇÃO CIENTÍFICA, Campinas, SP. Instituto de Zootecnia.

Wulfsohn, D., Zamora, F. A., Téllez, C. P., Lagos, I. Z., and García-Fiñana, M. (2012). Multilevel systematic sampling to estimate total fruit number for yield forecasts. Precision Agriculture, 13:256–275.