1907.11117.md - hassony2/inria-research-wiki GitHub Wiki
{paper} {notes}
Learning Visual Actions Using Multiple Verb-Only Labels, BMVC'19Michael Wray, Dima Damen
Motivation
Find a good representation of visual actions that is object-agnostic.
Explore soft and hard verb assignments for action recognition and retrieval.
Method
Types of verbs: - result verbs - close - open - sprinkle
- manner verbs
- pull
- hold
- touch