Jianghang Lin

MoHallBench: A Benchmark for Motion Hallucination in Video Large Language Models

MoHallBench is presented, a benchmark for diagnosing motion hallucination in VideoLLMs that systematically evaluates three major sources of hallucination: co-occurrence priors, sequential inference, and similarity confusion, and confirms that stronger priors and finer-grained similarity substantially amplify hallucination.