Counting Events, Not Seeing Them: Diagnosing and Repairing the Readout Failure in Video LLMs
Video LLMs fail conspicuously at counting how many times an object crosses the screen, a failure attributed to missing object permanence. Using construction-grounded synthetic testbeds with independently validated labels, we show this attribution is incomplete: the count signal is present but…
By Dongwu QianarXiv - cs.CV (CV)
Read the full story
Originally published by arXiv - cs.CV (CV) on Nov 10, 2026.
Comments
Sign in to join the discussion.