MVU-Eval: Towards Multi-Video Understanding Evaluation for Multimodal LLMs
It addresses the problem of evaluating multi-video understanding for researchers and developers in fields like autonomous driving and sports analytics, but it is incremental as it extends existing single-video benchmarks.
The paper tackles the lack of evaluation benchmarks for multi-video understanding in multimodal LLMs by introducing MVU-Eval, a comprehensive benchmark with 1,824 question-answer pairs across 4,959 videos, revealing significant performance gaps in current models.
The advent of Multimodal Large Language Models (MLLMs) has expanded AI capabilities to visual modalities, yet existing evaluation benchmarks remain limited to single-video understanding, overlooking the critical need for multi-video understanding in real-world scenarios (e.g., sports analytics and autonomous driving). To address this significant gap, we introduce MVU-Eval, the first comprehensive benchmark for evaluating Multi-Video Understanding for MLLMs. Specifically, our MVU-Eval mainly assesses eight core competencies through 1,824 meticulously curated question-answer pairs spanning 4,959 videos from diverse domains, addressing both fundamental perception tasks and high-order reasoning tasks. These capabilities are rigorously aligned with real-world applications such as multi-sensor synthesis in autonomous systems and cross-angle sports analytics. Through extensive evaluation of state-of-the-art open-source and closed-source models, we reveal significant performance discrepancies and limitations in current MLLMs' ability to perform understanding across multiple videos. The benchmark will be made publicly available to foster future research.