REDES CONVOLUCIONALES Y TRANSFORMERS PARA CLASIFICAR LA PUDRICIÓN NEGRA DEL CACAO: AUDITORÍA DE IDENTIDAD DE IMÁGENES Y SENSIBILIDAD AL CONTEXTO DE CAPTURA
DOI:
https://doi.org/10.31413/nat.v14i4.21893Palavras-chave:
Phytophthora spp., aprendizaje profundo, transferencia de aprendizaje, redes neuronales convolucionales, validación cruzada por grupos, OomicetosResumo
Convolutional networks and transformers for classifying cacao black pod rot: image identity audit and sensitivity to capture context
ABSTRACT: Black pod rot (BPR), caused by oomycetes of the genus Phytophthora, is one of the most destructive diseases of cacao. We compared five ImageNet-pretrained architectures (Swin-T, ConvNeXt-T, DeiT-Ti, EfficientNetV2-S, and ResNet50) for classifying healthy and BPR-affected pods in 2,211 images from a public repository. Before training, image identity was audited with perceptual hashes, DINOv2 descriptors, SIFT correspondences, and visual review; shots of the same pod or scene, present in 79% of the images, were merged into 664 groups. The models were trained with a common protocol and evaluated with group-stratified cross-validation (three repetitions of five folds), DeLong and McNemar tests, and Holm correction. Mean AUC ranged from 0.989 to 0.993 and accuracy from 0.954 to 0.964. Swin-T had the highest sensitivity and DeiT-Ti the lowest computational cost, but AUC differences were significant only in some comparisons and repetitions. On 1,958 unused healthy images, specificity (0.952 to 0.965) was similar to that of cross-validation, except for a capture scene that appeared only in diseased pods during training, where it dropped to 0.52 to 0.85. The five architectures discriminate well and similarly, but they also learn context associations, so field use requires external validation with images of both classes taken in the same contexts.
Keywords: Phytophthora spp.; deep learning; transfer learning; convolutional neural networks; group cross-validation; Oomycetes.
Referências
ABADE, A.; FERREIRA, P. A.; VIDAL, F. de B. Plant diseases recognition on images using convolutional neural networks: a systematic review. Computers and Electronics in Agriculture, v. 185, e106125, 2021. https://doi.org/10.1016/j.compag.2021.106125
ADEBAYO, J.; GILMER, J.; MUELLY, M.; GOODFELLOW, I.; HARDT, M.; KIM, B. Sanity checks for saliency maps. In: CONFERENCE ON NEURAL INFORMATION PROCESSING SYSTEMS, 32, 2018, Montreal. Advances in Neural Information Processing Systems, v. 31, p. 9505-9515, 2018. Disponible en: https://proceedings.neurips.cc/paper/2018/hash/294a8ed24b1ad22ec2e7efea049b8737-Abstract.html
AHMAD, A.; GAMAL, A. E.; SARASWAT, D. Toward generalization of deep learning-based plant disease identification under controlled and field conditions. IEEE Access, v. 11, p. 9042-9057, 2023. https://doi.org/10.1109/ACCESS.2023.3240100
BARBEDO, J. G. A. Deep learning applied to plant pathology: the problem of data representativeness. Tropical Plant Pathology, v. 47, n. 1, p. 85-94, 2022. https://doi.org/10.1007/s40858-021-00459-9
BARBEDO, J. G. A. Impact of dataset size and variety on the effectiveness of deep learning and transfer learning for plant disease classification. Computers and Electronics in Agriculture, v. 153, p. 46-53, 2018. https://doi.org/10.1016/j.compag.2018.08.013
BATES, S.; HASTIE, T.; TIBSHIRANI, R. Cross-validation: what does it estimate and how well does it do it? Journal of the American Statistical Association, v. 119, n. 546, p. 1434-1445, 2024. https://doi.org/10.1080/01621459.2023.2197686
BEAKES, G. W.; GLOCKLING, S. L.; SEKIMOTO, S. The evolutionary phylogeny of the oomycete "fungi". Protoplasma, v. 249, n. 1, p. 3-19, 2012. https://doi.org/10.1007/s00709-011-0269-2
BRIER, G. W. Verification of forecasts expressed in terms of probability. Monthly Weather Review, v. 78, n. 1, p. 1-3, 1950. https://doi.org/10.1175/1520-0493(1950)078<0001:VOFEIT>2.0.CO;2
DELONG, E. R.; DELONG, D. M.; CLARKE-PEARSON, D. L. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics, v. 44, n. 3, p. 837-845, 1988. https://doi.org/10.2307/2531595
DENG, J.; DONG, W.; SOCHER, R.; LI, L.-J.; LI, K.; FEI-FEI, L. ImageNet: a large-scale hierarchical image database. In: IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, 2009, Miami. Proceedings [...]. Piscataway: IEEE, 2009. p. 248-255. https://doi.org/10.1109/CVPR.2009.5206848
DIETTERICH, T. G. Approximate statistical tests for comparing supervised classification learning algorithms. Neural Computation, v. 10, n. 7, p. 1895-1923, 1998. https://doi.org/10.1162/089976698300017197
FAWCETT, T. An introduction to ROC analysis. Pattern Recognition Letters, v. 27, n. 8, p. 861-874, 2006. https://doi.org/10.1016/j.patrec.2005.10.010
FISCHLER, M. A.; BOLLES, R. C. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Communications of the ACM, v. 24, n. 6, p. 381-395, 1981. https://doi.org/10.1145/358669.358692
GADO, G. A. S.; PRIMANDARI, P. N. Sistem klasifikasi berbasis Android untuk penyakit buah kakao menggunakan CNN NasNet-Mobile. Jurnal Teknologi Terpadu, v. 11, n. 1, p. 27-35, 2025. https://doi.org/10.54914/jtt.v11i1.1821
GEIRHOS, R.; JACOBSEN, J.-H.; MICHAELIS, C.; ZEMEL, R.; BRENDEL, W.; BETHGE, M.; WICHMANN, F. A. Shortcut learning in deep neural networks. Nature Machine Intelligence, v. 2, n. 11, p. 665-673, 2020. https://doi.org/10.1038/s42256-020-00257-z
GODMALIN, R. A.; ALIAC, C. J.; FELISCUZO, L. Cacao pod infection level classification using transfer learning. In: IEEE OPEN CONFERENCE OF ELECTRICAL, ELECTRONIC AND INFORMATION SCIENCES (eStream), 2023, Vilnius. Proceedings [...]. Piscataway: IEEE, 2023. 6p. https://doi.org/10.1109/eStream59056.2023.10135062
GOODFELLOW, I.; BENGIO, Y.; COURVILLE, A. Deep learning. Cambridge: MIT Press, 2016. 800p.
GUEST, D. Black pod: diverse pathogens with a global impact on cocoa yield. Phytopathology, v. 97, n. 12, p. 1650-1653, 2007. https://doi.org/10.1094/PHYTO-97-12-1650
GUO, C.; PLEISS, G.; SUN, Y.; WEINBERGER, K. Q. On calibration of modern neural networks. In: INTERNATIONAL CONFERENCE ON MACHINE LEARNING, 34, 2017, Sydney. Proceedings of Machine Learning Research [...]. v. 70, p. 1321-1330, 2017. Disponible en: https://proceedings.mlr.press/v70/guo17a.html
HE, K.; ZHANG, X.; REN, S.; SUN, J. Deep residual learning for image recognition. In: IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, 2016, Las Vegas. Proceedings [...]. Piscataway: IEEE, 2016. p. 770-778. https://doi.org/10.1109/CVPR.2016.90
HOLM, S. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, v. 6, n. 2, p. 65-70, 1979. Disponible en: https://www.jstor.org/stable/4615733
HOWARD, J.; RUDER, S. Universal language model fine-tuning for text classification. In: ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS, 56, 2018, Melbourne. Proceedings [...]. Stroudsburg: ACL, 2018. p. 328-339. https://doi.org/10.18653/v1/P18-1031
IOFFE, S.; SZEGEDY, C. Batch normalization: accelerating deep network training by reducing internal covariate shift. In: INTERNATIONAL CONFERENCE ON MACHINE LEARNING, 32, 2015, Lille. Proceedings of Machine Learning Research, v. 37, p. 448-456, 2015. Disponible en: https://proceedings.mlr.press/v37/ioffe15.html
KAMOUN, S.; FUZER, O.; JONES, J. D. G.; JUDELSON, H. S.; ALI, G. S.; DALIO, R. J. D.; ROY, S. G.; SCHENA, L.; ZAMBOUNIS, A.; et al. The Top 10 oomycete pathogens in molecular plant pathology. Molecular Plant Pathology, v. 16, n. 4, p. 413-434, 2015. https://doi.org/10.1111/mpp.12190
KAPOOR, S.; NARAYANAN, A. Leakage and the reproducibility crisis in machine-learning-based science. Patterns, v. 4, n. 9, e100804, 2023. https://doi.org/10.1016/j.patter.2023.100804
KORNBLITH, S.; SHLENS, J.; LE, Q. V. Do better ImageNet models transfer better? In: IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, 2019, Long Beach. Proceedings [...]. Piscataway: IEEE, 2019. p. 2661-2671. https://doi.org/10.1109/CVPR.2019.00277
LIU, Z.; LIN, Y.; CAO, Y.; HU, H.; WEI, Y.; ZHANG, Z.; LIN, S.; GUO, B. Swin Transformer: hierarchical vision transformer using shifted windows. In: IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION, 2021, Montreal. Proceedings [...]. Piscataway: IEEE, 2021. p. 9992-10002. https://doi.org/10.1109/ICCV48922.2021.00986
LIU, Z.; MAO, H.; WU, C.-Y.; FEICHTENHOFER, C.; DARRELL, T.; XIE, S.; A ConvNet for the 2020s. In: IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, 2022, New Orleans. Proceedings [...]. Piscataway: IEEE, 2022. p. 11966-11976. https://doi.org/10.1109/CVPR52688.2022.01167
LOSHCHILOV, I.; HUTTER, F. Decoupled weight decay regularization. In: INTERNATIONAL CONFERENCE ON LEARNING REPRESENTATIONS, 7, 2019, New Orleans. Proceedings [...]. 2019. Disponible en: https://openreview.net/forum?id=Bkg6RiCqY7
LOWE, D. G. Distinctive image features from scale-invariant keypoints. International Journal of Computer Vision, v. 60, n. 2, p. 91-110, 2004. https://doi.org/10.1023/B:VISI.0000029664.99615.94
LUQUE, A.; CARRASCO, A.; MARTÍN, A.; LAS HERAS, A. The impact of class imbalance in classification performance metrics based on the binary confusion matrix. Pattern Recognition, v. 91, p. 216-231, 2019. https://doi.org/10.1016/j.patcog.2019.02.023
MARELLI, J.-P.; GUEST, D. I.; BAILEY, B. A.; EVANS, H. C.; BROWN, J. K.; JUNAID, M.; BARRETO, R. W.; LISBOA, D. O.; PUIG, A. S. Chocolate under threat from old and new cacao diseases. Phytopathology, v. 109, n. 8, p. 1331-1343, 2019. https://doi.org/10.1094/PHYTO-12-18-0477-RVW
MAYLIANTI, N. P.; WIJAYAKUSUMA, I. G. N. L.; WIGUNA, I. P. C. A. Comparison of EfficientNet-B0 and ResNet-50 for detecting diseases in cocoa fruit. Journal of Applied Informatics and Computing, v. 9, n. 1, p. 115-120, 2025. https://doi.org/10.30871/jaic.v9i1.8868
MCNEMAR, Q. Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika, v. 12, n. 2, p. 153-157, 1947. https://doi.org/10.1007/BF02295996
MOHANTY, S. P.; HUGHES, D. P.; SALATHÉ, M. Using deep learning for image-based plant disease detection. Frontiers in Plant Science, v. 7, e1419, 2016. https://doi.org/10.3389/fpls.2016.01419
NAVARRO-CABRERA, J. R.; BERAÚN-BARRANTES, J. G.; CÁRDENAS-GARCÍA, Á.; LOZANO-CARRANZA, C. M. Aplicación de aprendizaje profundo en la detección fitosanitaria del cacao usando visión por computadora. Revista Científica de Sistemas e Informática, v. 6, n. 1, e1385, 2026. https://doi.org/10.51252/rcsi.v6i1.1385
OQUAB, M.; DARCET, T.; MOUTAKANNI, T.; VO, H. V.; SZAFRANIEC, M.; KHALIDOV, V.; FERNANDEZ, P.; HAZIZA, D.; MASSA, F.; EL-NOUBY, A.; et al. DINOv2: learning robust visual features without supervision. Transactions on Machine Learning Research, p. 1-32, 2024. Disponible en: https://openreview.net/forum?id=a68SUt6zFt
PAGADUAN, Z. Jr. Cacao Diseases. Kaggle, 2021. Conjunto de datos. Disponible en: https://www.kaggle.com/datasets/zaldyjr/cacao-diseases
PASZKE, A.; GROSS, S.; MASSA, F.; LERER, A.; BRADBURY, J.; CHANAN, G.; KILLEEN, T.; LIN, Z.; GIMELSHEIN, N.; ANTIGA, L.; et al. PyTorch: an imperative style, high-performance deep learning library. In: CONFERENCE ON NEURAL INFORMATION PROCESSING SYSTEMS, 33, 2019, Vancouver. Advances in Neural Information Processing Systems, v. 32, p. 8024-8035, 2019. Disponible en: https://papers.nips.cc/paper/9015-pytorch-an-imperative-style-high-performance-deep-learning-library
PERRINE-WALKER, F. Phytophthora palmivora-cocoa interaction. Journal of Fungi, v. 6, n. 3, 167, 2020. https://doi.org/10.3390/jof6030167
PRADANA, A. G.; SETIADI, D. R. I. M.; MUSLIKH, A. R. Fine tuning model convolutional neural network EfficientNet-B4 dengan augmentasi data untuk klasifikasi penyakit kakao. Journal of Information System and Application Development, v. 2, n. 1, p. 1-11, 2024. https://doi.org/10.26905/jisad.v2i1.11899
RECHT, B.; ROELOFS, R.; SCHMIDT, L.; SHANKAR, V. Do ImageNet classifiers generalize to ImageNet? In: INTERNATIONAL CONFERENCE ON MACHINE LEARNING, 36, 2019, Long Beach. Proceedings of Machine Learning Research, v. 97, p. 5389-5400, 2019. Disponible en: https://proceedings.mlr.press/v97/recht19a.html
ROLA, J. B.; BARRERA, J. J. A.; CALHOUN, M. V.; ORAÑO-MAAGHOP, J. F.; UNAJAN, M. C.; BONCALON, J. M.; SEBIOS, E. T.; ESPINOSA, J. S. Convolutional neural network model for cacao Phytophthora palmivora disease recognition. International Journal of Advanced Computer Science and Applications, v. 15, n. 8, e97, 2024. https://doi.org/10.14569/IJACSA.2024.0150897
SELVARAJU, R. R.; COGSWELL, M.; DAS, A.; VEDANTAM, R.; PARIKH, D.; BATRA, D. Grad-CAM: visual explanations from deep networks via gradient-based localization. In: IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION, 2017, Venice. Proceedings [...]. Piscataway: IEEE, 2017. p. 618-626. https://doi.org/10.1109/ICCV.2017.74
SUDJUD, S.; JAMIL, M.; MELATI, R. Development of cocoa pod rot disease identification model using You Only Look Once (YOLO-v9). Mathematical Modelling of Engineering Problems, v. 12, n. 4, p. 1268-1274, 2025. https://doi.org/10.18280/mmep.120418
SUN, X.; XU, W. Fast implementation of DeLong's algorithm for comparing the areas under correlated receiver operating characteristic curves. IEEE Signal Processing Letters, v. 21, n. 11, p. 1389-1393, 2014. https://doi.org/10.1109/LSP.2014.2337313
TAN, D. S.; LEONG, R. N.; LAGUNA, A. F.; NGO, C. A.; LAO, A.; AMALIN, D. M.; ALVINDIA, D. G. AuToDiDAC: automated tool for disease detection and assessment for cacao black pod rot. Crop Protection, v. 103, p. 98-102, 2018. https://doi.org/10.1016/j.cropro.2017.09.017
TAN, M.; LE, Q. EfficientNetV2: smaller models and faster training. In: INTERNATIONAL CONFERENCE ON MACHINE LEARNING, 38, 2021, virtual. Proceedings of Machine Learning Research, v. 139, p. 10096-10106, 2021. Disponible en: https://proceedings.mlr.press/v139/tan21a.html
TECHIE-MENSON, H.; ASANTE, M.; MISSAH, Y. M.; ABDUL-SALAAM, G.; OPPONG, S. O. Enhanced convolutional block attention module with learnable gated fusion (LGF-CBAM) for cocoa pod disease identification. PLOS ONE, v. 21, n. 4, e0348147, 2026. https://doi.org/10.1371/journal.pone.0348147
TOUVRON, H.; CORD, M.; DOUZE, M.; MASSA, F.; SABLAYROLLES, A.; JEGOU, H. Training data-efficient image transformers & distillation through attention. In: INTERNATIONAL CONFERENCE ON MACHINE LEARNING, 38, 2021, virtual. Proceedings of Machine Learning Research, v. 139, p. 10347-10357, 2021. Disponible en: https://proceedings.mlr.press/v139/touvron21a.html
VALERO-GÓMEZ, J. C.; CLARES-PERCA, J. C.; ZÚÑIGA-INCALLA, A. P. Redes convolucionales y Transformers para clasificar la pudrición negra del cacao (Phytophthora spp.): código y datos derivados. Versión 1.0.1. [S. l.]: Zenodo, 2026. Programa de computador. DOI: https://doi.org/10.5281/zenodo.23178066
VILLALOBOS-CULQUI, C.; GARCÍA-RIVAS-PLATA, C.; TUESTA-HIDALGO, O. A. Artificial vision model based on convolutional neural networks for black pod identification in cacao plantations. Revista Científica de Sistemas e Informática, v. 5, n. 1, e678, 2025. https://doi.org/10.51252/rcsi.v5i1.678
WIGHTMAN, R. PyTorch Image Models. GitHub, 2019. Repositorio de software. Disponible en: https://github.com/huggingface/pytorch-image-models
YANAC-MONTESINO, R.; ROSALES-HUAMANI, J. A.; CASTILLO-SEQUERA, J. L. Detection of Phytophthora palmivora in cocoa fruit with deep learning. In: IBERIAN CONFERENCE ON INFORMATION SYSTEMS AND TECHNOLOGIES (CISTI), 16, 2021, Chaves. Proceedings [...]. Piscataway: IEEE, 2021. p. 1-4. https://doi.org/10.23919/CISTI52073.2021.9476279
YOSINSKI, J.; CLUNE, J.; BENGIO, Y.; LIPSON, H. How transferable are features in deep neural networks? In: CONFERENCE ON NEURAL INFORMATION PROCESSING SYSTEMS, 28, 2014, Montreal. Advances in Neural Information Processing Systems, v. 27, p. 3320-3328, 2014. Disponible en: https://proceedings.neurips.cc/paper_files/paper/2014/hash/532a2f85b6977104bc93f8580abbb330-Abstract.html
ZECH, J. R.; BADGELEY, M. A.; LIU, M.; COSTA, A. B.; TITANO, J. J.; OERMANN, E. K. Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: a cross-sectional study. PLOS Medicine, v. 15, n. 11, e1002683, 2018. https://doi.org/10.1371/journal.pmed.1002683
Downloads
Publicado
Edição
Seção
Como Citar
Licença
Copyright (c) 2026 Juan Carlos Valero-Gómez, Juan Carlos Clare- Perca , Alex Peter Zúñiga-Incalla
Este trabalho está licenciado sob uma licença Creative Commons Attribution-NonCommercial 4.0 International License.
Direitos Autorais para artigos publicados nesta revista são do autor, com direitos de primeira publicação para a revista. Em virtude de a aparecerem nesta revista de acesso público, os artigos são de uso gratuito, com atribuições próprias, em aplicações educacionais e não-comerciais.
A artigos publicados nessa revista, podem ser reproduzidos parcialmente ou utilizados como referência por outros autores, desde que seja cita a fonte, ou seja, a Revista Nativa.
Copyright for articles published in this journal are the authors, with first publication rights granted to the journal. The journal shows open access, and articles are free to use, with proper attribution, in educational and non-commercial.
The articles published in this journal may be reproduced in part or used as a reference by other authors, provided that the source is quoted.
