[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-72364":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":23,"hasPages":23,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":32,"readmeContent":33,"aiSummary":34,"trendingCount":16,"starSnapshotCount":16,"syncStatus":35,"lastSyncTime":36,"discoverSource":37},72364,"VideoRAG","HKUDS\u002FVideoRAG","HKUDS","[KDD'2026] \"VideoRAG: Chat with Your Videos\"","https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.01549",null,"Python",3053,431,52,18,0,12,29,62,36,29.91,"Other",false,"main",[26,27,28,29,30,31],"large-language-models","llms","long-video-understanding","multi-modal-llms","rag","retrieval-augmented-generation","2026-06-12 02:03:02","\u003Cdiv align=\"center\">\n  \u003Cpicture>\n      \u003Cimg src=\"cover.png\" width=\"80%\" style=\"border: none; box-shadow: none;\" alt=\"Vimo: Chat with Your Videos\">\n  \u003C\u002Fpicture>\n  \n  \u003Ch1>\n    \u003Cstrong>VideoRAG: Chat with Your Videos\u003C\u002Fstrong> • \u003Cstrong>Vimo Desktop\u003C\u002Fstrong>\n  \u003C\u002Fh1>\n\n  \u003Ca href=\"https:\u002F\u002Ftrendshift.io\u002Frepositories\u002F16146\" target=\"_blank\">\u003Cimg src=\"https:\u002F\u002Ftrendshift.io\u002Fapi\u002Fbadge\u002Frepositories\u002F16146\" alt=\"HKUDS%2FVideoRAG | Trendshift\" style=\"width: 250px; height: 55px;\" width=\"250\" height=\"55\"\u002F>\u003C\u002Fa>\n  \n  \u003Ca href='https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.01549'>\u003Cimg src='https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FarXiv-2502.01549-b31b1b'>\u003C\u002Fa>\n  \u003Ca href='https:\u002F\u002Fgithub.com\u002FHKUDS\u002FVideoRAG\u002Fissues\u002F1'>\u003Cimg src='https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F群聊-wechat\u002Ffeishu-green'>\u003C\u002Fa>\n  \u003Ca href='https:\u002F\u002Fdiscord.gg\u002FZzU55kz3'>\u003Cimg src='https:\u002F\u002Fdiscordapp.com\u002Fapi\u002Fguilds\u002F1296348098003734629\u002Fwidget.png?style=shield'>\u003C\u002Fa>\n  \u003Ca href='https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=D5vsxcp4QZI'>\u003Cimg src='https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-Watch%20Demo-red?style=flat&logo=youtube'>\u003C\u002Fa>\n  [![Blog](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FBlog-LearnOpenCV-blue?style=flat&logo=data:image\u002Fpng;base64,iVBORw0KGgoAAAANSUhEUgAAAJYAAACWCAMAAAAL34HQAAAAilBMVEVHcEwuLi4qKio3NzdQUFBjY2NoaGhiYmJ8fHx4eHiGhoabm5t1dXW2tranp6eOjo7Pz8\u002F\u002F\u002F\u002F\u002F9\u002Ff35+fn29vbz8\u002FPv7+\u002Ft7e3q6urn5+fj4+Pg4OA4svIyrfAsp+0noesinekemOcalOUchMVjZGQXaZxOT08OT3oMPFwvMTIONEoKIS8OFx0DAwPBWB\u002F1AAAAEXRSTlMACxw5ZXqQmqG1vdfv8ff7\u002FXwvPHUAAAnaSURBVHja3ZyJkqI6GIUbXHpcwHSUbtuMdtuiLcu8\u002F+vdCIRDjElYxLLuz701W03NV+ccAmT5XzqV4w6Go\u002FFkPl\u002F4RS3m88l4NBy4zsuDC0Sv05nnEcLYhvH\u002FebHsp4R43mz6+ng2ZzD8M+NAOcaGI6Hy38ngZn+Gg4eRuYPR1LsQVWk+iwtVoE1HA\u002FcROo1nORPLcW4WBwSaNxv3rJk7nGRMGwlpfV2CrSDbEG8ydPsTajQjFZXWJc\u002FH+gMFuAteqdls1I9k3D1Sca4AQr3zyn4EXa6bAOvBS2fw6mdQYCpheAWVKyuwZaIVYP7rfcHcUqkKkwC6UQWeICsl44q595NqOCUQCkwFwuq6gJaTQbINBxs6dwrVxMuUUphAtFwti1ot+a8A914hK7z0JoN7+Dea5VLlUIJpFeQ8uip0WwkyDgbBRm53qWSokglItKg3KqpkA5kM1lEwZzgjhX+AqjDREgklACtkAMuc7Jowd+wJqQAlmMBzq3I2QQYwIZjX\u002FpYcTAkT\u002FmVQEMqIBDReQrIMDE4yMm1pZGkgoCDUW80CGcAuXGzDjWwVK5+xMlWZfRDKjqOC5VZCMMb8odOcymOsTBWkAlMzslIwJIwxrymXM\u002FJgoALVFQxcI6cR1VihAlQXMImLlzd2GlMVBopUAaotmJQwnlpwNXDQKFV3wYRe8NGadpUKUF3qysiMq27uVSpIdRfBwFXcj7Wo\u002FAqVRqqugkl6Mb8G12CWj1f9UMFI5Iux2cBG5U4J02jVHxeZurahgfSmFbjeJK5PzjV2LHHXUPWslzc0B0s8BzlVAKoe9BK5z8d7Y7zc7A25ORWly+DyD\u002FAnQu2\u002FUXCJYYJM9PEaIVh1qehqTfxFGJ6irE5huPDJekXr6gWukcVCBItSGxPxwyhN0yQ+FxUn\u002FJdR6BMbGaWIl9FGBxbWolqyxSlKk\u002FPv8XD4+fnZ\u002F+z3\u002FIfD4fh7TtLotGDLmlzCRsdwF66FhRaqFZlzJo70c6M4Giebk5WZCzbq70Z3imBZqQISRskZTLfIzkkUksDOVdo4dTUDqWShwT4ScqHApCPjkoXEYOW1jWNHm3d7sOjHIkoBZQZLo8UHtcZLm3rnNRfLbiElpzQGlKWOcXoi1G5jLtercy2WX9PCwI\u002FgX536TSI\u002FqGmjP1CTVSvv6\u002FlFqv1PgzrE6XxdJ\u002FVIF5IliaWjoixMz6pUe7kUrnMaMqrjglxqukYQy2QhO6W\u002FGiQ9GNf2Nz0xk42QaySNWRDLYCHlVMfbUN9l3SY7ci6qt1HIxdjMrQ7wuA0NYilUQEIBTeEyyIWbcVh9GmLM0ou1DmWqK6av7JLIZK5wbZQrd5FNHATew5ilFSuYy7kC1NdVAUwaKNJ5oJULY5eH0IvRwXQbUj89a6DAZQQ7pz413YxijFACbxCLkig+qFSCYycKZArXIY4INcglxghX8VAv1scp0VJdaPiV\u002F8eropjMlZzWerngIgYtW+CXi2qwAAWdpJIEq8ZrsbSEHkOXM7V7KCxUqQTXtrgA9q1ycRvtLk6duh4GYSpTyVJtqyUpJrgwSgR1XRzKHt7MOyxUqS4ogKqgcSrBBRs9nVxwcZh5+Mfq4Sqs5D2nunBdKfWXX5cSZn5Br0rqw5XVxT8OhgeThyRSxZKp\u002FqK2kl6KXBGxujhzRbSMT+nlXIgFKkBlTIASYLutFC\u002FINV\u002Fq70WEyx4tyiKM79AKVGqVgsFGyMVqheuVWKJFF+nxplgyFMfYKlyqjcd0QW\u002FKhXCRV4xa+mitTvBQthBUh2Oc8IqPh62GCy6eVtpwYeRC4hEtNfAQq7QQWu2OcfqvqDQ+7gQXbIRcCL0hXDM3T7wxWj48hFg80YLqEP+TKj6AC3LBxZsvEteZH3oi8StgqYOWilVSJf+uKgFXRS5p6DJn3hviOa1L\u002FDo6yx5CLFDJlR5kuWQXz9H6Ta1sQMXTekwsiceDRxYLVGol4FJd\u002FE2JJfNk\u002FDJhFiwpWhArp9oiV1LFWe4hVwVrz8NlwWKTl7nu0YNR66B6KCw8\u002FtPUUWCp9+IBI5fu8TN\u002FmW\u002FM4wMNEwULHsY6rHgLFwUWMn8Ta1libeYvCwnrRhTDWI1WmfcUIJrU3whXHC7NI8Rm8eIzM1ZwOktYiod2F3fXWKfgJhZM9DmWeTR9j4Bl9xAVq5nHCPGuVas9FkzcJnqspBqub+lWPEcfdixupWmQX19hfVdM3JmwdsiWota6o1oq1ldzLFUtYKEUtYxYH7fU2tYx0aDWR2e1Ak227JHXZ8se+f4HCHXcOp8CG9YzDacYtxZ9P3x2moePeZSf9\u002FuohloNH9WtXmyAtbO+2Kg3Yq0XmzFhbV4Dd+1fA481XgNbvzRvDVyJ8aU51gzy0kszPjGCOp8Y+KQ2f2KoHlpmR+RPjBYfZMCyfZDtWn+Qtf983dk\u002FX7cNP18xZdPxY\u002F+v6WN\u002F1\u002F5jXz81grrn1MgeUyNKtDA1UmuOUjuRZJ5H6jSRhMxrw0XnycE0GajHajvtdqdJShkJWrWdpKw3pbtUp3QFlzynK5hA1XxKl2VTuoZwocwT4CBTqNpNgNdcLqDm5YILh4Sko9ofk3rLBd0XV+TaVplaL67UXoqi+qWoL4WJU+06LkWpLi7rLdxh3RVkQLrHwp3TeZlzl7Hxy7DMua+9zNl1URhg+kXhfeNF4WZL6Cv9Evp3myV0lLKEfr8NB9hvAChIVXvDwYZNnIbbM+iDt2d038zyvecsF572m1kQeJRj2foDLuQLZGope7iglVLK1p+eNkr9dN4ohaq5rYzm28rU2udX921lLTfhrRpuwju02YSHqr1lcdloy+Kh5ZbF\u002Fjd4rhts8Oy0HZY+YDssxq4mm4e9MMKGZt225ij0mmwenrlPu9W6+8Z0io3pYOu6Mb37Nn7+h5238cPCzWzwDIceqHLo4RmOiLypR0Se4EANrXGgBuU86PjRm3r86AkOa1GJCgOpvtxp71yqVmTqPsVBwJWgWuMgYKNjk8FDjk0+3SFTxP1JjuSCyml6BL3\u002FA8xNDqI7Y3DdTzBACSoc936Gw\u002FFBcTj+c8NA9RStBODgczZeePY2Fd2betCuTT1Ed5b\u002FSwsUNIxhbRvGUEPDGIaGMQ9qr0PL0rfX2XRt4OQMJkQIBjCQAa1aS6nlD6CEVGQy6Kl1U6C0bspYUCswaVo39d\u002FoSuHpt9EV2oIVYCBDozK5JVi1LxiEQr8ypOrxTdT4b9uaqPXfck5uOlcUWs4BCi3nHtKgL4MoWN4\u002FQASmDRr03b2kdoY5GuBQ6LX4iHaGaP4IMgH3WW2tiOaPn4IJzR\u002F7AuNeVttSantlPqhVJsqRGosKPFwbILFHNBYFGNqwSnQA0rRhfXDTWtTmPk1rn7TF739gu3see8j9YQAAAABJRU5ErkJggg==)](https:\u002F\u002Flearnopencv.com\u002Fvideorag-long-context-video-comprehension\u002F)\n  [![Platform](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fplatform-macOS%20|%20Windows%20|%20Linux-lightgrey.svg)]()\n  \n\n  **🎬 Intelligent Video Conversations | Powered by Advanced AI | Extreme Long-Context Processing**\n\n\u003C\u002Fdiv>\n\n\u003Cbr\u002F>\n\n\u003Cimg src='VideoRAG-algorithm\u002FVideoRAG_cover.png' \u002F>\n\nVimo is a revolutionary desktop application that lets you **chat with your videos** using cutting-edge AI technology. Built on the powerful [VideoRAG framework](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.01549), Vimo can understand and analyze videos of any length - from short clips to hundreds of hours of content - and answer your questions with remarkable accuracy.\n\n### 🎥 Watch Vimo in Action\n\nSee how Vimo transforms video interaction with intelligent conversations and deep understanding capabilities.\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=D5vsxcp4QZI\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.youtube.com\u002Fvi\u002FD5vsxcp4QZI\u002Fmaxresdefault.jpg\" width=\"80%\" alt=\"Vimo Introduction Video\">\n  \u003C\u002Fa>\n  \u003Cp>\u003Cem>👆 Click to watch the Vimo demo video\u003C\u002Fem>\u003C\u002Fp>\n\u003C\u002Fdiv>\n\n## ✨ Key Features\n\n### For Everyone\n- **Drag & Drop Upload**: Simply drag video files into Vimo\n- **Smart Conversations**: Ask questions in natural language\n- **Multi-Format Support**: Works with MP4, MKV, AVI, and more\n- **Cross-Platform**: Available on macOS, Windows, and Linux\n\n### For Power Users\n- **Extreme Long Videos**: Process videos up to hundreds of hours\n- **Multi-Video Analysis**: Compare and analyze multiple videos simultaneously\n- **Advanced Retrieval**: Find specific moments and scenes with precision\n- **Export Capabilities**: Save insights and references for later use\n\n### For Researchers\n- **VideoRAG Framework**: Access to cutting-edge retrieval-augmented generation\n- **Benchmark Dataset**: LongerVideos benchmark with 134+ hours of content\n- **Performance Metrics**: Detailed evaluation against existing methods\n- **Extensible Architecture**: Build upon our open-source foundation\n  \n## 🌟 Why Vimo?\n\n**For Video Enthusiasts & Professionals:**\n- **Effortless Video Analysis**: Upload any video and start asking questions immediately\n- **Natural Conversations**: Chat with your videos as if talking to a human expert\n- **No Length Limits**: Process everything from 30-second clips to 100+ hour documentaries\n- **Deep Understanding**: Combines visual content, audio, and context for comprehensive answers\n\n**For Researchers & Developers:**\n- **State-of-the-Art Algorithm**: Built on VideoRAG, featuring graph-driven knowledge indexing\n- **Benchmark Performance**: Evaluated on 134+ hours across lectures, documentaries, and entertainment\n- **Open Source**: Full access to VideoRAG implementation and research findings\n- **Scalable Architecture**: Efficient processing with single GPU (RTX 3090) capability\n\n## 📋 Table of Contents\n\n- [🚀 Quick Start](#-quick-start)\n- [✨ Key Features](#-key-features)\n- [🔬 VideoRAG Algorithm](#-videorag-algorithm)\n- [🛠️ Development Setup](#️-development-setup)\n- [🧪 Benchmarks & Evaluation](#-benchmarks--evaluation)\n- [📖 Citation](#-citation)\n- [🤝 Contributing](#-contributing)\n- [🙏 Acknowledgement](#-acknowledgement)\n\n## 🚀 Quick Start of Vimo\n\n### Option 1: Download Vimo App (Coming Soon)\n\n> [!NOTE]\n> We are preparing the **Beta release** for macOS Apple Silicon first, with Windows and Linux versions coming soon!\n\n\u003Cdiv align=\"left\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FHKUDS\u002FVimo\u002Freleases\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FComing%20Soon-Mac%20Download-007ACC?style=for-the-badge&logo=apple&logoColor=white\" alt=\"Coming Soon - Mac Release\" height=\"50\">\n  \u003C\u002Fa>\n\u003C\u002Fdiv>\n\n### Option 2: Run from Source Code\n\nFor detailed setup instructions:\n\n- **Vimo Desktop App**: See [Vimo-desktop](Vimo-desktop) for complete installation and configuration steps\n\n**Quick Overview:**\n1. Set up the Python backend environment and start the VideoRAG server\n2. Launch the Electron frontend application\n3. Start chatting with your videos!\n\n## 🔬 VideoRAG Algorithm\n\n\u003Cp align=\"center\">\n\u003Cimg src=\"VideoRAG-algorithm\u002FVideoRAG.png\" alt=\"VideoRAG Architecture\" width=\"80%\" \u002F>\n\u003C\u002Fp>\n\nVideoRAG introduces a novel dual-channel architecture that combines:\n\n- **Graph-Driven Knowledge Indexing**: Multi-modal knowledge graphs for structured video understanding\n- **Hierarchical Context Encoding**: Preserves spatiotemporal visual patterns across long sequences  \n- **Adaptive Retrieval**: Dynamic retrieval mechanisms optimized for video content\n- **Cross-Video Understanding**: Semantic relationship modeling across multiple videos\n\n### Technical Highlights\n\n- **Efficient Processing**: Handle hundreds of hours on a single RTX 3090 (24GB)\n- **Structured Indexing**: Distill long videos into concise knowledge representations\n- **Multi-Modal Retrieval**: Align textual queries with visual and audio content\n- **LongerVideos Benchmark**: 160+ videos, 134+ hours across diverse domains\n\n### Performance Comparison\n\nOur VideoRAG algorithm significantly outperforms existing methods in long-context video understanding:\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"Vimo-desktop\u002Ffigures\u002Ftable.png\" width=\"80%\" alt=\"Performance Comparison\" \u002F>\n\u003C\u002Fdiv>\n\nWe also evaluate VideoRAG's QA performance on the Video-MME long video track to better understand the gains over the backbone models (included here because of the paper's page limit):\n\n| Video-MME Long Video | MiniCPM-o w\u002Fo subs | MiniCPM-o w\u002F subs | MiniCPM-V w\u002Fo subs | MiniCPM-V w\u002F subs | VideoRAG |\n| --- | ---: | ---: | ---: | ---: | ---: |\n| Accuracy | 52.2% | 56.3% | 51.8% | 56.3% | **60.2%** |\n\n> Note: The score may show slight fluctuations across runs due to the instability of LLM generation.\n\n### Experiments and Evaluation\n\nSee [VideoRAG-algorithm](VideoRAG-algorithm) for detailed development setup including:\n- Conda environment creation\n- Model checkpoints download\n- Dependencies installation\n- Evaluation scripts\n\n## 🧪 LongerVideos Benchmark\n\nWe created the LongerVideos benchmark to evaluate long-context video understanding:\n\n| Video Type       | #Collections | #Videos | #Queries | Avg. Duration |\n|------------------|-------------|---------|----------|---------------|\n| **Lectures**     | 12          | 135     | 376      | ~64.3 hours   |\n| **Documentaries**| 5           | 12      | 114      | ~28.5 hours   |\n| **Entertainment**| 5           | 17      | 112      | ~41.9 hours   |\n| **Total**        | 22          | 164     | 602      | ~134.6 hours  |\n\nFor detailed evaluation instructions and reproduction scripts, see [VideoRAG-algorithm\u002Freproduce](VideoRAG-algorithm\u002Freproduce).\n\n## 📖 Citation\n\nIf you find Vimo or VideoRAG helpful in your research, please cite our paper:\n\n```bibtex\n@article{VideoRAG,\n  title={VideoRAG: Retrieval-Augmented Generation with Extreme Long-Context Videos},\n  author={Ren, Xubin and Xu, Lingrui and Xia, Long and Wang, Shuaiqiang and Yin, Dawei and Huang, Chao},\n  journal={arXiv preprint arXiv:2502.01549},\n  year={2025}\n}\n```\n\n## 🤝 Contributing\n\nWe welcome contributions from the community! Whether you're:\n\n- **Reporting bugs** or suggesting features for Vimo\n- **Improving VideoRAG algorithms** or adding new capabilities  \n- **Enhancing documentation** or creating tutorials\n- **Designing UI\u002FUX improvements** for better user experience\n\nFeel free to submit issues and pull requests. Together, we're building the future of intelligent video interaction!\n\n## 🙏 Acknowledgement\n\nVimo builds upon the incredible work of the open-source community:\n\n- **[VideoRAG](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.01549)**: The core algorithm powering Vimo's intelligence\n- **[nano-graphrag](https:\u002F\u002Fgithub.com\u002Fgusye1234\u002Fnano-graphrag)** & **[LightRAG](https:\u002F\u002Fgithub.com\u002FHKUDS\u002FLightRAG)**: Graph-based retrieval foundations\n- **[ImageBind](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FImageBind)**: Multi-modal representation learning\n- **[uitars-desktop](https:\u002F\u002Fgithub.com\u002Fbytedance\u002FUI-TARS-desktop)**: Desktop application architecture inspiration\n\n**🌟 Transform how you interact with videos. Start your journey with Vimo today!**\n\n---\n\n\u003Cdiv align=\"center\">\n  \u003Csub>Built with ❤️ by the VideoRAG@HKUDS team.\u003C\u002Fsub>\n\u003C\u002Fdiv> \n","VideoRAG 是一个基于大语言模型的视频理解和对话系统，允许用户通过自然语言与视频内容进行交互。该项目的核心功能包括长视频理解、多模态大语言模型以及检索增强生成技术，能够从视频中提取关键信息并以对话形式提供给用户。它特别适用于需要对视频内容进行深入分析和互动的场景，如教育、娱乐、新闻报道等领域。项目采用 Python 语言开发，已在 GitHub 上获得了广泛的关注。",2,"2026-06-11 03:41:32","high_star"]