Dendrobot: Phytosanitary Diagnosis of Rubber Tree Tapping Panels via Semantic Segmentation with Deep Learning
Resumo
Brazilian rubber plantations lose up to 30% of latex productivity due to undetected infectious diseases in tapping panels. This paper presents Dendrobot, a semantic segmentation system based on U-Net with ResNet34 backbone, capable of automatically diagnosing three infectious diseases — Black Crust, Dry Crust (TPD) and Pink Mold — and generating agronomic prescriptions from a single field photograph. A dataset of 727 images was manually annotated with six semantic classes in COCO JSON format and partitioned via stratified 3-way split (588 train / 55 validation / 55 test). A hybrid Dice + Categorical Focal Loss with 35:1 class weighting overcomes background collapse; a Hierarchical Thresholding Engine eliminates minority-class suppression. The final model achieves 83.59% global accuracy and 71.33% mean IoU: Black Crust (61.48%), Dry Crust (71.48%), Pink Mold (81.04%). Functional accuracy (mAP@0.25) reaches 93.3%.Referências
Bernardes, M. S. (2005). Sangria da seringueira. EPAMIG, Viçosa, MG.
Chen, L.-C., Zhu, Y., Papandreou, G., Schroff, F., and Adam, H. (2018). Encoder-decoder with atrous separable convolution for semantic image segmentation. In European Conference on Computer Vision (ECCV), pages 801–818.
Kerkech, M., Hafiane, A., and Canals, R. (2020). Deep learning approach with colorimetric spaces and multiple vegetation indices for vine diseases detection in UAV images. Computers and Electronics in Agriculture, 155:237–243.
Li, H. and Ye, L. (2023). Tapping panel dryness detection based on Snake Optimizer combined with Otsu thresholding. Computers and Electronics in Agriculture, 204:107523.
Li, H., Zhang, C., and Luo, W. (2018). Identification of tapping panel dryness of Hevea brasiliensis using image processing based on Otsu’s method. In Proc. International Conference on Image and Video Processing, and Artificial Intelligence.
Lin, T.-Y., Goyal, P., Girshick, R., He, K., and Dollár, P. (2017). Focal loss for dense object detection. In IEEE International Conference on Computer Vision (ICCV), pages 2980–2988.
Milletari, F., Navab, N., and Ahmadi, S.-A. (2016). V-net: Fully convolutional neural networks for volumetric medical image segmentation. In Fourth International Conference on 3D Vision (3DV), pages 565–571.
Roboflow (2024). Roboflow: Computer vision tools for annotation, training, and deployment. [link]. Accessed: June 2026.
Ronneberger, O., Fischer, P., and Brox, T. (2015). U-Net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Intervention (MICCAI), volume 9351 of Lecture Notes in Computer Science, pages 234–241. Springer.
Ultralytics (2023). YOLOv8: A new state-of-the-art computer vision model. [link].
Xie, E., Wang, W., Yu, Z., Anandkumar, A., Alvarez, J. M., and Luo, P. (2021). SegFormer: Simple and efficient design for semantic segmentation with transformers. In Advances in Neural Information Processing Systems (NeurIPS), volume 34, pages 12077–12090.
Xie, Y., Liang, X., and Chen, J. (2024). DBO-Otsu: A multi-threshold segmentation method for rubber tree tapping panel images. Biosystems Engineering, 238:1–12.
Chen, L.-C., Zhu, Y., Papandreou, G., Schroff, F., and Adam, H. (2018). Encoder-decoder with atrous separable convolution for semantic image segmentation. In European Conference on Computer Vision (ECCV), pages 801–818.
Kerkech, M., Hafiane, A., and Canals, R. (2020). Deep learning approach with colorimetric spaces and multiple vegetation indices for vine diseases detection in UAV images. Computers and Electronics in Agriculture, 155:237–243.
Li, H. and Ye, L. (2023). Tapping panel dryness detection based on Snake Optimizer combined with Otsu thresholding. Computers and Electronics in Agriculture, 204:107523.
Li, H., Zhang, C., and Luo, W. (2018). Identification of tapping panel dryness of Hevea brasiliensis using image processing based on Otsu’s method. In Proc. International Conference on Image and Video Processing, and Artificial Intelligence.
Lin, T.-Y., Goyal, P., Girshick, R., He, K., and Dollár, P. (2017). Focal loss for dense object detection. In IEEE International Conference on Computer Vision (ICCV), pages 2980–2988.
Milletari, F., Navab, N., and Ahmadi, S.-A. (2016). V-net: Fully convolutional neural networks for volumetric medical image segmentation. In Fourth International Conference on 3D Vision (3DV), pages 565–571.
Roboflow (2024). Roboflow: Computer vision tools for annotation, training, and deployment. [link]. Accessed: June 2026.
Ronneberger, O., Fischer, P., and Brox, T. (2015). U-Net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Intervention (MICCAI), volume 9351 of Lecture Notes in Computer Science, pages 234–241. Springer.
Ultralytics (2023). YOLOv8: A new state-of-the-art computer vision model. [link].
Xie, E., Wang, W., Yu, Z., Anandkumar, A., Alvarez, J. M., and Luo, P. (2021). SegFormer: Simple and efficient design for semantic segmentation with transformers. In Advances in Neural Information Processing Systems (NeurIPS), volume 34, pages 12077–12090.
Xie, Y., Liang, X., and Chen, J. (2024). DBO-Otsu: A multi-threshold segmentation method for rubber tree tapping panel images. Biosystems Engineering, 238:1–12.
Publicado
19/10/2026
Como Citar
LOPES, José Vitor de Amorim; SILVA, Sérgio Francisco da.
Dendrobot: Phytosanitary Diagnosis of Rubber Tree Tapping Panels via Semantic Segmentation with Deep Learning. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 23. , 2026, Cuiabá/MT.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 2955-2963.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2026.31033.
