CinemaCLIP: A hybrid CLIP model for the visual language of cinema

What's the use case for this and who are the customers?