[ECCV 2026] VKnowU: Evaluating Visual Knowledge Understanding in Multimodal LLMs - View it on GitHub