Bridging Modalities, Spanning Time: Structured Memory for Ultra-Long Agentic Video Reasoning

TL;DR

MAGIC-Video introduces a multimodal memory graph and narrative chain to enable long-range reasoning over ultra-long videos, improving retrieval and understanding across days or weeks.

Contribution

It presents a training-free framework that unifies episodic, semantic, and visual content for ultra-long video reasoning, outperforming existing baselines.

Findings

Outperforms strong baselines on EgoLifeQA, Ego-R1, and MM-Lifelong benchmarks.

Achieves 10.1, 7.4, and 5.9 point improvements over prior systems.

Effectively handles modality and time dimensions in ultra-long video understanding.

Abstract

Understanding ultra-long videos such as egocentric recordings, live streams, or surveillance footage spanning days to weeks, remains a challenge. For current multimodal LLMs: even with million-token context windows, frame budgets cover only tens of minutes of densely sampled video, and most evidence is discarded before inference begins. Memory-augmented and agentic approaches help with scale, but their retrieval remains fragmented across modalities and lacks long-range narrative summaries that span days or weeks. We propose \textbf{MAGIC-Video}, a training-free framework built around a multimodal memory graph with interleaved narrative chain: the graph unifies episodic, semantic, and visual content through six typed edges and supports cross-modal retrieval, while the chain distils long-horizon entity biographies and recurring activity events. At inference time, an agentic loop interleaves graph retrieval with narrative fact injection, covering both the modality and time dimensions of ultra-long video in a single retrieval pipeline. On EgoLifeQA, Ego-R1 and MM-Lifelong, MAGIC-Video consistently outperforms strong general-purpose, long-video, and agentic baselines, with gains of 10.1, 7.4, and 5.9 points over the prior best agentic system on each benchmark. Code is available at https://github.com/lijiazheng0917/MAGIC-video.