When building an Identify-Binder AI model in GeoDict-AI, a large number of structure models created in GrainGeo are used as a training dataset.
In the Other hand, is it possible to use an image model in which particles and binders have already been identified as a training dataset instead of this?