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

2026-05-23

AI Summary

This study addresses the challenge of mining footprint segmentation in remote sensing imagery, where scarce fine-grained annotations and domain shift between coarse- and fine-labeled data hinder performance. To tackle this, the authors propose MineC2FNet, a novel framework that introduces coarse-to-fine domain incremental learning to this task for the first time. Built upon a teacher–student architecture, MineC2FNet employs attention-based distillation mechanisms at both feature and prediction levels to transfer generalizable knowledge from abundant coarse-boundary data while leveraging limited fine annotations to refine boundary details. The work also contributes a new dataset comprising 219 high-resolution, multi-regional images with precise annotations. Experimental results demonstrate that the proposed method significantly outperforms existing domain adaptation and domain incremental learning approaches across multiple metrics. The code and dataset are publicly released.

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.

Problem

• mining footprint segmentation
• domain shift
• coarse-to-fine learning
• data scarcity
• multispectral imagery

Innovation

• coarse-to-fine learning
• domain incremental learning
• attentive distillation
• mining footprint segmentation
• multispectral imagery