不同材料成像条件需匹配不同分割架构，本文提供选型指南与可信度评估。
Context Determines Optimal Architecture in Materials Segmentation
- 构建跨模态评价框架，覆盖SEM、AFM、XCT等四类成像
- 高对比度2D图像用UNet，复杂情况首选DeepLabv3+
- 可检测分布外样本并解释预测依据，助研究者信任模型
分割架构通常在单一成像模态上进行评测，掩盖了实际部署中的性能差异：某一模态最优的架构在另一模态上可能表现不佳。本文提出一个涵盖SEM、AFM、XCT和光学显微镜的跨模态材料图像分割评价框架。对六种编码器-解码器组合在七个数据集上的评估表明，最优架构随成像上下文系统性变化：在高对比度2D成像中，UNet表现最佳；而在最复杂的场景下，DeepLabv3+更优。该框架还通过分布外检测和反事实解释提供部署反馈，揭示驱动预测的关键微结构特征。整体为材料表征提供了架构选择指导、可靠性信号与可解释性工具，填补了研究者缺乏针对特定成像设置选型及判断模型可信度的实践空白。
原文摘要 · Abstract (English)
Segmentation architectures are typically benchmarked on single imaging modalities, obscuring deployment-relevant performance variations: an architecture optimal for one modality may underperform on another. We present a cross-modal evaluation framework for materials image segmentation spanning SEM, AFM, XCT, and optical microscopy. Our evaluation of six encoder-decoder combinations across seven datasets reveals that optimal architectures vary systematically by context: UNet excels for high-contrast 2D imaging while DeepLabv3+ is preferred for the hardest cases. The framework also provides deployment feedback via out-of-distribution detection and counterfactual explanations that reveal which microstructural features drive predictions. Together, the architecture guidance, reliability signals, and interpretability tools address a practical gap in materials characterization, where researchers lack tools to select architectures for their specific imaging setup or assess when models can be trusted on new samples.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。