📄 中文摘要
该项目旨在根据用户输入生成高度特定的扫描电子显微镜（SEM）晶圆图像。研究团队发现，通过分割来表达用户意图是最直观和准确的方法。分割是一种简单的结构布局，能够有效地表示用户希望生成的微观结构。为了支持这一方法，团队需要高质量的配对数据集，包括 SEM 图像及其对应的分割掩码。由于缺乏这样的数据集，项目的初期阶段面临挑战，但通过与行业合作伙伴的协作，逐步克服了数据收集和处理的问题。
从分割掩码生成 SEM 图像
出处: Generating SEM Images from Segmentation Masks
发布: 2026年2月17日
📄 English Summary
Generating SEM Images from Segmentation Masks
The project aimed to generate highly specific scanning electron microscope (SEM) wafer images based on user input. The research team discovered that segmentation is the most intuitive and accurate way for users to express their intent. Segmentation serves as a simple structural layout that effectively represents the micro-structure users wish to generate. To support this approach, the team required a high-quality paired dataset of SEM images and their corresponding segmentation masks. The initial phase of the project faced challenges due to the lack of such datasets, but through collaboration with industry partners, the team gradually overcame issues related to data collection and processing.