# SceneRAG: Scene-level Retrieval-Augmented Generation for Video Understanding

arXiv: https://arxiv.org/abs/2506.07600 · HTML: https://mlsanity.com/paper/2506.07600/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal RAG, one of 15 directions in Retrieval-augmented generation.
- Builds on 15 of its 21 references that are in the map.
- First posted on arXiv 9 Jun 2025.

## Placed in

- Retrieval-augmented generation / [Multimodal RAG](https://mlsanity.com/rag/multimodal-rag/)

## Builds on (most cited first, 15 of 15)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Retrieval-Augmented Generation for Large Language Models: A Survey](https://mlsanity.com/paper/2312.10997/) (2023)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [From Local to Global: A Graph RAG Approach to Query-Focused Summarization](https://mlsanity.com/paper/2404.16130/) (2024)
- [MiniCPM-V: A GPT-4V Level MLLM on Your Phone](https://mlsanity.com/paper/2408.01800/) (2024)
- [VideoAgent: Long-form Video Understanding with Large Language Model as Agent](https://mlsanity.com/paper/2403.10517/) (2024)
- [LightRAG: Simple and Fast Retrieval-Augmented Generation](https://mlsanity.com/paper/2410.05779/) (2024)
- [Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling](https://mlsanity.com/paper/2311.00430/) (2023)
- [VideoRAG: Retrieval-Augmented Generation with Extreme Long-Context Videos](https://mlsanity.com/paper/2502.01549/) (2025)
- [VideoRAG: Retrieval-Augmented Generation over Video Corpus](https://mlsanity.com/paper/2501.05874/) (2025)
- [Retrieval-Augmented Egocentric Video Captioning](https://mlsanity.com/paper/2401.00789/) (2024)
- [Multi-granularity Correspondence Learning from Long-term Noisy Videos](https://mlsanity.com/paper/2401.16702/) (2024)
- [Understanding Long Videos via LLM-Powered Entity Relation Graphs](https://mlsanity.com/paper/2501.15953/) (2025)
- [RAVU: Retrieval Augmented Video Understanding with Compositional Reasoning over Graph](https://mlsanity.com/paper/2505.03173/) (2025)
- [Scene Graph Generation Strategy with Co-occurrence Knowledge and Learnable Term Frequency](https://mlsanity.com/paper/2405.12648/) (2024)

## Built on by (most cited first, 7 of 7)

- [VideoStir: Understanding Long Videos via Spatio-Temporally Structured and Intent-Aware RAG](https://mlsanity.com/paper/2604.05418/) (2026)
- [Roles of MLLMs in Visually Rich Document Retrieval for RAG: A Survey](https://mlsanity.com/paper/2601.03262/) (2025)
- [Spatiotemporal Knowledge Graphs as Persistent Scene Memory for Embodied Question Answering](https://mlsanity.com/paper/2510.01483/) (2025)
- [CRAFT: Critic-Refined Adaptive Key-Frame Targeting for Multimodal Video Question Answering](https://mlsanity.com/paper/2605.19075/) (2026)
- [Event-Causal RAG: A Retrieval-Augmented Generation Framework for Long Video Reasoning in Complex Scenarios](https://mlsanity.com/paper/2605.06185/) (2026)
- [Beyond Visual Boundaries: Rethinking Scene Segmentation for Movie RAG](https://mlsanity.com/paper/2608.28699/) (2026)
- [A Simulation-Grounded Agentic VLM Framework for Wildfire Monitoring and Reporting](https://mlsanity.com/paper/2610.02451/) (2026)

## Abstract (arXiv)

Despite recent advances in retrieval-augmented generation (RAG) for video understanding, effectively understanding long-form video content remains underexplored due to the vast scale and high complexity of video data. Current RAG approaches typically segment videos into fixed-length chunks, which often disrupts the continuity of contextual information and fails to capture authentic scene boundaries. Inspired by the human ability to naturally organize continuous experiences into coherent scenes, we present SceneRAG, a unified framework that leverages large language models to segment videos into narrative-consistent scenes by processing ASR transcripts alongside temporal metadata. SceneRAG further sharpens these initial boundaries through lightweight heuristics and iterative correction. For each scene, the framework fuses information from both visual and textual modalities to extract entity relations and dynamically builds a knowledge graph, enabling robust multi-hop retrieval and generation that account for long-range dependencies. Experiments on the LongerVideos benchmark, featuring over 134 hours of diverse content, confirm that SceneRAG substantially outperforms prior baselines, achieving a win rate of up to 72.5 percent on generation tasks.