Coarse-to-Fine Domain Incremental Learning with Attentive Distillation for Mining Footprint Segmentation in Multispectral Imagery

Handoyo, Alif Tri; Lee, Vincent C.S.; Purwanto, Rizka Widyarini; Lechner, Alex M.; Kemp, Deanna; Saputra, Muhamad Risqi U.

doi:10.24963/ijcai.2026/800

Coarse-to-Fine Domain Incremental Learning with Attentive Distillation for Mining Footprint Segmentation in Multispectral Imagery

Alif Tri Handoyo, Vincent C.S. Lee, Rizka Widyarini Purwanto, Alex M. Lechner, Deanna Kemp, Muhamad Risqi U. Saputra

Download BibTeX Cite

Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence. AI and Social Good. Pages 7192-7200.

https://doi.org/10.24963/ijcai.2026/800

Keywords

AI4G: Multidisciplinary Topics and Applications
AI4G: Computer Vision
AI4G: Machine Learning

Abstract

Automatically mapping and segmenting global mining footprints using remote sensing and deep learning is critical for monitoring the socio-environmental risks and impacts of mining, yet its progress is hindered by the scarcity of fine-grained annotated data. Although large-scale datasets with coarse boundaries are widely available, leveraging them to improve fine-grained segmentation is challenging due to significant domain shift. To address this, we propose MineC2FNet, a coarse-to-fine domain incremental learning framework that exploits abundant coarse data to enhance fine-grained mining footprint segmentation. MineC2FNet adopts a teacher–student architecture with attentive distillation at both the feature and prediction levels, selectively transferring generalized knowledge from the coarse domain while enabling boundary refinement using limited fine-grained data (fine domain). We further introduce an expertly validated dataset of 219 images with precise boundary annotations across diverse geographies and commodities. Extensive experiments against state-of-the-art approaches, including domain adaptation and domain incremental learning methods, demonstrate that MineC2FNet achieves superior performance while effectively handling domain shift. The dataset and code are publicly available at https://github.com/risqiutama/MineC2FNet.

Alif Tri Handoyo, Vincent C.S. Lee, Rizka Widyarini Purwanto, Alex M. Lechner, Deanna Kemp, and Muhamad Risqi U. Saputra. 2026. Coarse-to-Fine Domain Incremental Learning with Attentive Distillation for Mining Footprint Segmentation in Multispectral Imagery. In Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-26. IJCAI, 7192–7200. https://doi.org/10.24963/ijcai.2026/800

@inproceedings{ijcai2026p800,
  title     = {Coarse-to-Fine Domain Incremental Learning with Attentive Distillation for Mining Footprint Segmentation in Multispectral Imagery},
  author    = {Handoyo, Alif Tri and Lee, Vincent C.S. and Purwanto, Rizka Widyarini and Lechner, Alex M. and Kemp, Deanna and Saputra, Muhamad Risqi U.},
  booktitle = {Proceedings of the Thirty-Fifth International Joint Conference on
               Artificial Intelligence, {IJCAI-26}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {Diego Calvanese},
  pages     = {7192--7200},
  year      = {2026},
  month     = {8},
  note      = {AI and Social Good},
  doi       = {10.24963/ijcai.2026/800},
  url       = {https://doi.org/10.24963/ijcai.2026/800},
}

TY  - CPAPER
TI  - Coarse-to-Fine Domain Incremental Learning with Attentive Distillation for Mining Footprint Segmentation in Multispectral Imagery
AU  - Handoyo, Alif Tri
AU  - Lee, Vincent C.S.
AU  - Purwanto, Rizka Widyarini
AU  - Lechner, Alex M.
AU  - Kemp, Deanna
AU  - Saputra, Muhamad Risqi U.
T2  - Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-26
PB  - International Joint Conferences on Artificial Intelligence Organization
PY  - 2026
SP  - 7192
EP  - 7200
DO  - 10.24963/ijcai.2026/800
UR  - https://doi.org/10.24963/ijcai.2026/800
ER  -