[KongchangAI]
ConceptVFM

Vision Foundation Model

A foundation model pretrained on large-scale visual data that possesses general-purpose visual representation capabilities and can be transferred to a variety of downstream vision tasks.

Source Articles