MLLMs drop from over 85% accuracy on action presence to under 50% on matched action-denial videos, exposing a causal verification gap that causal graph prompts partially close.
Ross, João Carreira, Alexander V ostrikov, and Andrew Zisserman
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.CV 3verdicts
UNVERDICTED 3representative citing papers
Introduces the QEVD benchmark for asynchronous situated interaction in fitness coaching and proposes a streaming baseline to address limitations of existing vision-language models.
InternVideo combines masked video modeling and video-language contrastive learning into a single foundation model that reaches state-of-the-art results on 39 video datasets including 91.1% top-1 on Kinetics-400.
citing papers explorer
-
Learning to Deny: Action Denial in Multimodal Large Language Models
MLLMs drop from over 85% accuracy on action presence to under 50% on matched action-denial videos, exposing a causal verification gap that causal graph prompts partially close.
-
What to Say and When to Say it: Live Fitness Coaching as a Testbed for Situated Interaction
Introduces the QEVD benchmark for asynchronous situated interaction in fitness coaching and proposes a streaming baseline to address limitations of existing vision-language models.
-
InternVideo: General Video Foundation Models via Generative and Discriminative Learning
InternVideo combines masked video modeling and video-language contrastive learning into a single foundation model that reaches state-of-the-art results on 39 video datasets including 91.1% top-1 on Kinetics-400.