Towards Multimodal Lifelong Understanding: A Dataset and Agentic Baseline

Guo Chen, Lidong Lu, Yicheng Liu +17

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…