Coarse Indexing, Fine Evidence: Decoupling Temporal Granularity in Long-Video RAG
2026-08-24 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors show that when searching through long videos to answer questions, using a fixed size for splitting the video can be inefficient. They propose a method called Density-Aware Graph Construction (DAGC) that groups similar parts of the video together to create a simpler index for quick searching. After locating relevant parts roughly, the system then looks more closely at the original detailed segments to answer questions accurately. Their approach speeds up processing while keeping the answers nearly as good as before, and it works with different video understanding models.
Retrieval-Augmented Generation (RAG)Temporal GranularityGraph ConstructionLong-Video UnderstandingVideo SegmentationDensity-Adaptive IndexingQuestion Answering (QA)Video RetrievalVisual RedundancyEnd-to-End Acceleration
Authors
Zhe Jin, Zhimin Lin, Bin Zheng, Junhua Fang, Huihua Yang
Abstract
Graph-based retrieval-augmented generation (RAG) provides a scalable paradigm for long-video understanding, but existing systems typically inherit a fixed temporal granularity from video segmentation when constructing their retrieval index. We argue that this design unnecessarily couples indexing granularity with evidence granularity: coarse representations can often suffice for locating relevant temporal regions, while fine-grained evidence remains important for downstream reasoning. We propose \textbf{Density-Aware Graph Construction (DAGC)}, a training-free approach that decouples a query-independent coarse retrieval index from the original fine-grained evidence space. DAGC constructs a compact, density-adaptive graph index by merging visually redundant neighboring chunks, while preserving mappings to the original temporal units. Retrieved coarse regions are subsequently expanded back to the original chunk granularity for fine-grained evidence refinement and answer generation. Experiments on MLVU, VideoMME, and LongVideoBench show that DAGC retains only about 40--50\% of the original graph nodes and achieves $1.3$--$1.7\times$ end-to-end wall-clock acceleration while preserving approximately 99\% of the original QA performance. The gains transfer across different LVLM backbones and video RAG pipelines, suggesting that long-video RAG need not maintain the same temporal granularity for indexing and evidence reasoning.