AdaVideoRAG: Omni-Contextual Adaptive Retrieval-Augmented Efficient Long Video Understanding

TL;DR

AdaVideoRAG introduces an adaptive retrieval framework for long video understanding, dynamically balancing efficiency and reasoning depth through query-aware retrieval schemes and hierarchical knowledge bases.

Contribution

It proposes a novel adaptive RAG framework with an intent classifier and multi-modal knowledge indexing for improved long-video comprehension.

Findings

Significantly improves efficiency and accuracy on long-video QA tasks.

Seamlessly integrates into existing MLLMs with lightweight APIs.

Establishes a new paradigm for adaptive retrieval in video analysis.

Abstract

Multimodal Large Language Models (MLLMs) perform well in video understanding but degrade on long videos due to fixed-length context and weak long-term dependency modeling. Retrieval-Augmented Generation (RAG) can expand knowledge dynamically, yet existing video RAG schemes adopt fixed retrieval paradigms that ignore query difficulty. This uniform design causes redundant computation and latency for simple queries, while coarse retrieval for complex, multi-hop reasoning can miss key information. Such single-step retrieval severely limits the trade-off between efficiency and cognitive depth. We propose AdaVideoRAG, an adaptive RAG framework for long-video understanding. A lightweight intent classifier dynamically selects suitable retrieval schemes according to query complexity from the simplest to the most sophisticated. We design an Omni-Knowledge Indexing module that extracts and organizes multi-modal information into three databases: (1) a text base built from clip captions, ASR, and OCR; (2) a visual base; and (3) a knowledge graph for deep semantic understanding. This supports hierarchical knowledge access, from naive retrieval to graph-based retrieval, balancing resource cost and reasoning ability. To evaluate deep understanding, we further construct the HiVU benchmark. Experiments show that AdaVideoRAG significantly improves both efficiency and accuracy on long-video QA tasks and can be seamlessly plugged into existing MLLMs through lightweight APIs, establishing a new paradigm for adaptive retrieval-augmented video analysis.