A2O: LLM-based Agentic Learning of Action-to-Object Features for Video Action Recognition.

Taiki Sekii, Fumiaki Sato

Browse the full ACL paper archive.