ChristyHu/CellOPC

CellOPC是一个用于单元和上下文感知掩模优化的大规模基准数据集。它基于45纳米技术节点的真实集成电路布局构建，每个样本围绕标准单元放置实例进行裁剪，以保留单元级层次结构和周围布局上下文。该数据集提供配对的输入布局/目标图像和优化掩模图像，用于学习不同上下文大小下的单元级掩模生成。数据集包含基于模型的光学邻近校正和逆光刻技术掩模类型，旨在评估标准单元身份、相邻几何形状和输入上下文大小如何影响掩模预测和光刻感知可打印性。CellOPC is a large-scale benchmark dataset for cell- and context-aware mask optimization. It is constructed from real integrated circuit layouts at the 45 nm technology node. Each sample is clipped around a standard-cell placement instance to preserve cell-level hierarchy and surrounding layout context. The dataset provides paired input layout/target images and optimized mask images for learning cell-wise mask generation under different context sizes. The dataset contains both model-based OPC and ILT mask types. It is intended to evaluate how standard-cell identity, neighboring geometries, and input context size affect mask prediction and lithography-aware printability.