A Comparison of Incremental Semantic Segmentation Algorithms for the Recognition of Agricultural Threats from UAV Images

• Lucas Oliveira Antonetti UNESP
• Gabriel F. D. Pereira UNESP
• Rafael Gomes Mantovani UTFPR
• Luiz F. S. Coletta USP
• André Luís Debiaso Rossi UNESP

DOI: https://doi.org/10.5753/eniac.2026.31234

Resumo

This study investigates two Incremental Semantic Segmentation (ISS) frameworks, ALIFE and SATS, using a data-scarce UAV imagery for agricultural threat recognition. Using a 5-fold cross-validation, the methods are evaluated through a three-stage protocol that incrementally introduces base categories, corn, and the weed Panicum. In this setup, both methods achieved comparable final performance (ALIFE 51.48% mIoU, SATS 51.38%), though SATS showed a clear advantage on Ground recognition, while differences on Corn and Panicum were comparable between methods, and ALIFE exhibited more stable training convergence. Notably, SATS experienced complete background suppression, a degradation absent in standard benchmarks.

Referências

Carnevalli, Silva e Souza, S. (2020). Estudo de deep learning para o reconhecimento de ervas invasoras na cultura do milho a partir de imagens de vants.

Cermelli, F., Mancini, M., Bulò, S. R., Ricci, E., and Caputo, B. (2020). Modeling the background for incremental learning in semantic segmentation.

Chin, R., Catal, C., and Kassahun, A. (2023). Plant disease detection using drones in precision agriculture. Precision Agriculture, 24(5):1663–1682.

Cong, W., Cong, Y., and Ren, Y. (2025). Lightweight class incremental semantic segmentation without catastrophic forgetting. IEEE Transactions on Image Processing, 34:4566–4579.

Gupta, S. K., Yadav, S. K., Soni, S. K., Shanker, U., and Singh, P. K. (2023). Multiclass weed identification using semantic segmentation: An automated approach for precision agriculture. Ecological Informatics, 78:102366.

Hasan, A. M., Sohel, F., Diepeveen, D., Laga, H., and Jones, M. G. K. (2021). A survey of deep learning techniques for weed detection from images. Computers and Electronics in Agriculture, 184:106067.

Jin, X., Bagavathiannan, M., McCullough, P. E., Chen, Y., and Yu, J. (2022). A deep learning-based method for classification, detection, and localization of weeds in turfgrass. Pest Management Science, 78(11):4809–4821.

Kalb, T. and Beyerer, J. (2022). Causes of catastrophic forgetting in class-incremental semantic segmentation.

Kalb, T., Roschani, M., Ruf, M., and Beyerer, J. (2021). Continual learning for class- and domain-incremental semantic segmentation. In 2021 IEEE Intelligent Vehicles Symposium (IV), pages 1345–1351.

Mo, Y., Wu, Y., Yang, X., Liu, F., and Liao, Y. (2022). Review the state-of-the-art technologies of semantic segmentation based on deep learning. Neurocomputing, 493:626–646.

Oh, Y., Baek, D., and Ham, B. (2022). ALIFE: Adaptive logit regularizer and feature replay for incremental semantic segmentation. In 36th Conference on Neural Information Processing Systems (NeurIPS 2022).

Parisi, G. I., Kemker, R., Part, J. L., Kanan, C., and Wermter, S. (2019). Continual lifelong learning with neural networks: A review. Neural Networks, 113:54–71.

Pereira, G. F. D. (2023). Investigação de algoritmos de aprendizado profundo para a segmentação semântica de plantas daninhas e culturas agrícolas usando imagens.

Pimentel, M. A. F., Clifton, D. A., Clifton, L., and Tarassenko, L. (2014). A review of novelty detection. Signal Processing, 99:215–249.

Qiu, Y., Shen, Y., Sun, Z., Zheng, Y., Chang, X., Zheng, W., and Wang, R. (2023). Sats: Self-attention transfer for continual semantic segmentation. Pattern Recognition, 138:109383.

Rizzardi, M. A., Fleck, N. G., Vidal, R. A., Merotto Jr., A., and Agostinetto, D. (2001). Competição por recursos do solo entre ervas daninhas e culturas. Ciência Rural, 31(4):707–714.

Shan, L., Wang, W., Lv, K., and Luo, B. (2022). Class-incremental learning for semantic segmentation in aerial imagery via distillation in all aspects. IEEE Transactions on Geoscience and Remote Sensing, 60:1–12.

Silva, M. G. (2023). Classificador open-world para o reconhecimento de ameaças em imagens aéreas de culturas agrícolas. Dissertação de mestrado, Universidade Estadual Paulista “Júlio de Mesquita Filho”, Faculdade de Engenharia, São João da Boa Vista.

Tasar, O., Tarabalka, Y., and Alliez, P. (2019). Incremental learning for semantic segmentation of large-scale remote sensing data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 12(9):3524–3537.

Van de Ven, G. M., Tuytelaars, T., and Tolias, A. S. (2022). Three types of incremental learning. Nature Machine Intelligence, 4(12):1185–1197.

Wang, L., Zhang, X., Su, H., and Zhu, J. (2024). A comprehensive survey of continual learning: Theory, method and application. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(8):5362–5383.

Yuan, B. and Zhao, D. (2024). A survey on continual semantic segmentation: Theory, challenge, method and application. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(12):10891–10910.

Zhai, Z., Martínez, J. F., Beltran, V., and Martínez, N. L. (2020). Decision support systems for agriculture 4.0: Survey and challenges. Computers and Electronics in Agriculture, 170:105256.

Zhou, D.-W., Wang, Q.-W., Qi, Z.-H., Ye, H.-J., Zhan, D.-C., and Liu, Z. (2024). Class-incremental learning: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, pages 1–20.