EdgeAttNet: Towards Barb-Aware Filament Segmentation

2025-09-02

AI Summary

Current Hα-based filament segmentation—especially of barbs—suffers from insufficient accuracy, limiting effective modeling of long-range dependencies and fine-scale structures, thereby hindering reliable chirality determination, a critical predictor of coronal mass ejection (CME) behavior. To address this, we propose an Edge-Aware Transformer-enhanced U-Net: learnable edge maps are linearly embedded into the Key and Query matrices of self-attention layers, explicitly injecting structural priors to improve boundary and barb localization while reducing parameter count. The model is trained end-to-end and achieves state-of-the-art performance on the MAGFILO dataset, significantly outperforming standard U-Net and its variants in both filament and barb segmentation accuracy. It also demonstrates faster inference speed and lower computational overhead, indicating strong potential for practical deployment in solar physics applications.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Learning on the Edge & Model Compression
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Accurate segmentation of solar filaments in H-alpha observations is critical for determining filament chirality, a key factor in the behavior of Coronal Mass Ejections (CMEs). However, existing methods often fail to capture fine-scale filament structures, particularly barbs, due to a limited ability to model long-range dependencies and spatial detail. We propose EdgeAttNet, a segmentation architecture built on a U-Net backbone by introducing a novel, learnable edge map derived directly from the input image. This edge map is incorporated into the model by linearly transforming the attention Key and Query matrices with the edge information, thereby guiding the self-attention mechanism at the network's bottleneck to more effectively capture filament boundaries and barbs. By explicitly integrating this structural prior into the attention computations, EdgeAttNet enhances spatial sensitivity and segmentation accuracy while reducing the number of trainable parameters. Trained end-to-end, EdgeAttNet outperforms U-Net and other U-Net-based transformer baselines on the MAGFILO dataset. It achieves higher segmentation accuracy and significantly better recognition of filament barbs, with faster inference performance suitable for practical deployment.

Problem

• Accurate segmentation of solar filaments in H-alpha observations
• Existing methods fail to capture fine-scale filament structures like barbs
• Limited ability to model long-range dependencies and spatial detail

Innovation

• EdgeAttNet uses learnable edge map for segmentation
• Integrates edge information into self-attention mechanism
• Enhances spatial sensitivity with fewer parameters