Dataset for "DeepMoiré: Revealing hidden layer-resolved atomic structures in twisted 2D materials from HAADF-STEM images"

This dataset provides the essential training data and model weights for the DeepMoiré framework, designed for automated, layer-resolved atomic structure determination of twisted 2D materials and moiré superlattices, utilizing twisted bilayer MoS2 as a representative example. Included Content: ResUnet++ Module: Datasets containing exclusively twisted structures, utilized specifically for segmentation model training. ResUNet++ model weights trained directly on the provided dataset. DRIT Module: Datasets used for unsupervised domain adaptation and cross-domain alignment. DRIT model weights for bridging simulated images and experimental HAADF-STEM images.