Towards Multimodal Lifelong Understanding: A Dataset and Agentic Baseline
This dataset and agentic baseline address the problem of multimodal lifelong understanding in natural, unscripted daily life for AI agents, providing a more realistic benchmark than existing densely concatenated video datasets.
This paper introduces MM-Lifelong, a 181.1-hour dataset of unscripted daily life video, structured across Day, Week, and Month scales. It identifies that current MLLMs suffer from a Working Memory Bottleneck and agentic baselines from Global Localization Collapse on this data, proposing the Recursive Multimodal Agent (ReMA) to address these issues.
While datasets for video understanding have scaled to hour-long durations, they typically consist of densely concatenated clips that differ from natural, unscripted daily life. To bridge this gap, we introduce MM-Lifelong, a dataset designed for Multimodal Lifelong Understanding. Comprising 181.1 hours of footage, it is structured across Day, Week, and Month scales to capture varying temporal densities. Extensive evaluations reveal two critical failure modes in current paradigms: end-to-end MLLMs suffer from a Working Memory Bottleneck due to context saturation, while representative agentic baselines experience Global Localization Collapse when navigating sparse, month-long timelines. To address this, we propose the Recursive Multimodal Agent (ReMA), which employs dynamic memory management to iteratively update a recursive belief state, significantly outperforming existing methods. Finally, we establish dataset splits designed to isolate temporal and domain biases, providing a rigorous foundation for future research in supervised learning and out-of-distribution generalization.