AINeutralarXiv – CS AI · 14h ago6/10
🧠
Coarse-to-Fine Domain Incremental Learning with Attentive Distillation for Mining Footprint Segmentation in Multispectral Imagery
Researchers introduce MineC2FNet, a deep learning framework that leverages abundant coarse-grained remote sensing data to improve fine-grained mining footprint segmentation in multispectral imagery. The approach uses domain incremental learning with attentive distillation to bridge the gap between coarse and fine datasets, addressing a critical gap in environmental monitoring of global mining operations.