4DP-QA
Emerging1papers using it
2026first seen
The '4DP-QA' dataset contains 400,000 samples focused on motion-related scene understanding to evaluate the ability of Vision Language Models to reason about 4D scenes while disentangling object and camera motion.