The framework is implemented using three versions of the Contrastive Language-Image Pre-training (CLIP) model based on the combination of 12 datasets created by seven users containing more than ...