Fetching the paper…

C2C: Component-to-Composition Learning for Zero-Shot Compositional Action Recognition · Around