BVB: Benchmarking Agentic Video Understanding via Programmatic Reconstruction in Blender
A benchmark that tests the ability of agentic video understanding by asking agents to reconstruct real-world videos as animated Blender scenes, and shows that programmatic reconstruction is a viable test of agentic video understanding, and that semantic retention remains the main challenge.
Y. Y. Tang, D. Shimada, Jia-Yue Meng et al.
· 0 citations