Skip to content

Author

Sarah Ostadabbas

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Position: Video LLMs Must Not Ignore the Pixel Dynamics in Plain Sight

It is argued that recent progress in video understanding is measured by benchmarks and protocols that can be solved without reliably perceiving spatiotemporal evidence, rewarding language-driven plausibility over video-grounded inference.

Shayda Moezzi, Umer Saleem, Andong Deng et al. · 0 citations