[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-open-mmlab--mmyolo":3,"tool-open-mmlab--mmyolo":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":79,"owner_twitter":76,"owner_website":80,"owner_url":81,"languages":82,"stars":101,"forks":102,"last_commit_at":103,"license":104,"difficulty_score":10,"env_os":105,"env_gpu":106,"env_ram":105,"env_deps":107,"category_tags":117,"github_topics":118,"view_count":23,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":131,"updated_at":132,"faqs":133,"releases":163},1300,"open-mmlab\u002Fmmyolo","mmyolo","OpenMMLab YOLO series toolbox and benchmark. Implemented RTMDet, RTMDet-Rotated,YOLOv5, YOLOv6, YOLOv7, YOLOv8,YOLOX, PPYOLOE, etc.","MMYOLO 是 OpenMMLab 推出的 YOLO 系列一站式开发套件，把 RTMDet、YOLOv5\u002F6\u002F7\u002F8、YOLOX、PP-YOLOE 等主流目标检测模型整合到同一套框架里，并提供完整训练、测试、部署脚本和基准数据。它解决了“模型多、接口杂、复现难”的痛点：只需改几行配置，就能在 COCO、旋转框或自定义数据集上快速切换不同 YOLO 版本，并一键导出 ONNX、TensorRT 等格式。  \n开发者、算法研究员和工业落地团队都能受益：想发论文可直接调用统一 benchmark 做公平对比；做项目时，可按精度或速度需求挑模型，再借助自动化脚本在几分钟内部署到边缘设备。  \n亮点在于“模块化+即插即用”：主干网络、Neck、Head、数据增强策略全部解耦，像搭积木一样组合；同时支持混合精度、多机多卡、模型蒸馏等高级技巧，让训练更快、显存更省。","\u003Cdiv align=\"center\">\n  \u003Cimg width=\"100%\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_28d28e0f87c2.png\"\u002F>\n  \u003Cdiv> \u003C\u002Fdiv>\n  \u003Cdiv align=\"center\">\n    \u003Cb>\u003Cfont size=\"5\">OpenMMLab website\u003C\u002Ffont>\u003C\u002Fb>\n    \u003Csup>\n      \u003Ca href=\"https:\u002F\u002Fopenmmlab.com\">\n        \u003Ci>\u003Cfont size=\"4\">HOT\u003C\u002Ffont>\u003C\u002Fi>\n      \u003C\u002Fa>\n    \u003C\u002Fsup>\n        \n    \u003Cb>\u003Cfont size=\"5\">OpenMMLab platform\u003C\u002Ffont>\u003C\u002Fb>\n    \u003Csup>\n      \u003Ca href=\"https:\u002F\u002Fplatform.openmmlab.com\">\n        \u003Ci>\u003Cfont size=\"4\">TRY IT OUT\u003C\u002Ffont>\u003C\u002Fi>\n      \u003C\u002Fa>\n    \u003C\u002Fsup>\n  \u003C\u002Fdiv>\n  \u003Cdiv> \u003C\u002Fdiv>\n\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fmmyolo)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fmmyolo)\n[![docs](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdocs-latest-blue)](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002F)\n[![deploy](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fworkflows\u002Fdeploy\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Factions)\n[![codecov](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fopen-mmlab\u002Fmmyolo\u002Fbranch\u002Fmain\u002Fgraph\u002Fbadge.svg)](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fopen-mmlab\u002Fmmyolo)\n[![license](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fopen-mmlab\u002Fmmyolo.svg)](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002FLICENSE)\n[![open issues](https:\u002F\u002Fisitmaintained.com\u002Fbadge\u002Fopen\u002Fopen-mmlab\u002Fmmyolo.svg)](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues)\n[![issue resolution](https:\u002F\u002Fisitmaintained.com\u002Fbadge\u002Fresolution\u002Fopen-mmlab\u002Fmmyolo.svg)](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues)\n\n[📘Documentation](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002F) |\n[🛠️Installation](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002Fget_started\u002Finstallation.html) |\n[👀Model Zoo](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002Fmodel_zoo.html) |\n[🆕Update News](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002Fnotes\u002Fchangelog.html) |\n[🤔Reporting Issues](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues\u002Fnew\u002Fchoose)\n\n\u003C\u002Fdiv>\n\n\u003Cdiv align=\"center\">\n\nEnglish | [简体中文](README_zh-CN.md)\n\n\u003C\u002Fdiv>\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fopenmmlab.medium.com\u002F\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_062337b0e5ec.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Fdiscord.com\u002Fchannels\u002F1037617289144569886\u002F1046608014234370059\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_6342e5371027.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002FOpenMMLab\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_04c3beda0b07.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fopenmmlab\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_204fe79b5a90.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Fspace.bilibili.com\u002F1293512903\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_8655b6233577.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Fwww.zhihu.com\u002Fpeople\u002Fopenmmlab\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_447c4737c11f.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>\n\n## 📄 Table of Contents\n\n- [🥳 🚀 What's New](#--whats-new-)\n  - [✨ Highlight](#-highlight-)\n- [📖 Introduction](#-introduction-)\n- [🛠️ Installation](#%EF%B8%8F-installation-)\n- [👨‍🏫 Tutorial](#-tutorial-)\n- [📊 Overview of Benchmark and Model Zoo](#-overview-of-benchmark-and-model-zoo-)\n- [❓ FAQ](#-faq-)\n- [🙌 Contributing](#-contributing-)\n- [🤝 Acknowledgement](#-acknowledgement-)\n- [🖊️ Citation](#️-citation-)\n- [🎫 License](#-license-)\n- [🏗️ Projects in OpenMMLab](#%EF%B8%8F-projects-in-openmmlab-)\n\n## 🥳 🚀 What's New [🔝](#-table-of-contents)\n\n💎 **v0.6.0** was released on 15\u002F8\u002F2023:\n\n- Support YOLOv5 instance segmentation\n- Support YOLOX-Pose based on MMPose\n- Add 15 minutes instance segmentation tutorial.\n- YOLOv5 supports using mask annotation to optimize bbox\n- Add Multi-scale training and testing docs\n\nFor release history and update details, please refer to [changelog](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002Fnotes\u002Fchangelog.html).\n\n### ✨ Highlight [🔝](#-table-of-contents)\n\nWe are excited to announce our latest work on real-time object recognition tasks, **RTMDet**, a family of fully convolutional single-stage detectors. RTMDet not only achieves the best parameter-accuracy trade-off on object detection from tiny to extra-large model sizes but also obtains new state-of-the-art performance on instance segmentation and rotated object detection tasks. Details can be found in the [technical report](https:\u002F\u002Farxiv.org\u002Fabs\u002F2212.07784). Pre-trained models are [here](configs\u002Frtmdet).\n\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frtmdet-an-empirical-study-of-designing-real\u002Freal-time-instance-segmentation-on-mscoco)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002Freal-time-instance-segmentation-on-mscoco?p=rtmdet-an-empirical-study-of-designing-real)\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frtmdet-an-empirical-study-of-designing-real\u002Fobject-detection-in-aerial-images-on-dota-1)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002Fobject-detection-in-aerial-images-on-dota-1?p=rtmdet-an-empirical-study-of-designing-real)\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frtmdet-an-empirical-study-of-designing-real\u002Fobject-detection-in-aerial-images-on-hrsc2016)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002Fobject-detection-in-aerial-images-on-hrsc2016?p=rtmdet-an-empirical-study-of-designing-real)\n\n| Task                     | Dataset | AP                                   | FPS(TRT FP16 BS1 3090) |\n| ------------------------ | ------- | ------------------------------------ | ---------------------- |\n| Object Detection         | COCO    | 52.8                                 | 322                    |\n| Instance Segmentation    | COCO    | 44.6                                 | 188                    |\n| Rotated Object Detection | DOTA    | 78.9(single-scale)\u002F81.3(multi-scale) | 121                    |\n\n\u003Cdiv align=center>\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_b5b09232c93f.png\"\u002F>\n\u003C\u002Fdiv>\n\nMMYOLO currently implements the object detection and rotated object detection algorithm, but it has a significant training acceleration compared to the MMDeteciton version. The training speed is 2.6 times faster than the previous version.\n\n## 📖 Introduction [🔝](#-table-of-contents)\n\nMMYOLO is an open source toolbox for YOLO series algorithms based on PyTorch and [MMDetection](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection). It is a part of the [OpenMMLab](https:\u002F\u002Fopenmmlab.com\u002F) project.\n\nThe master branch works with **PyTorch 1.6+**.\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_6ddac5574998.gif\"\u002F>\n\n\u003Cdetails open>\n\u003Csummary>Major features\u003C\u002Fsummary>\n\n- 🕹️ **Unified and convenient benchmark**\n\n  MMYOLO unifies the implementation of modules in various YOLO algorithms and provides a unified benchmark. Users can compare and analyze in a fair and convenient way.\n\n- 📚 **Rich and detailed documentation**\n\n  MMYOLO provides rich documentation for getting started, model deployment, advanced usages, and algorithm analysis, making it easy for users at different levels to get started and make extensions quickly.\n\n- 🧩 **Modular Design**\n\n  MMYOLO decomposes the framework into different components where users can easily customize a model by combining different modules with various training and testing strategies.\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_c5633eb648bb.jpg\" alt=\"BaseModule-P5\"\u002F>\n  The figure above is contributed by RangeKing@GitHub, thank you very much!\n\nAnd the figure of P6 model is in [model_design.md](docs\u002Fen\u002Frecommended_topics\u002Fmodel_design.md).\n\n\u003C\u002Fdetails>\n\n## 🛠️ Installation [🔝](#-table-of-contents)\n\nMMYOLO relies on PyTorch, MMCV, MMEngine, and MMDetection. Below are quick steps for installation. Please refer to the [Install Guide](docs\u002Fen\u002Fget_started\u002Finstallation.md) for more detailed instructions.\n\n```shell\nconda create -n mmyolo python=3.8 pytorch==1.10.1 torchvision==0.11.2 cudatoolkit=11.3 -c pytorch -y\nconda activate mmyolo\npip install openmim\nmim install \"mmengine>=0.6.0\"\nmim install \"mmcv>=2.0.0rc4,\u003C2.1.0\"\nmim install \"mmdet>=3.0.0,\u003C4.0.0\"\ngit clone https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo.git\ncd mmyolo\n# Install albumentations\npip install -r requirements\u002Falbu.txt\n# Install MMYOLO\nmim install -v -e .\n```\n\n## 👨‍🏫 Tutorial [🔝](#-table-of-contents)\n\nMMYOLO is based on MMDetection and adopts the same code structure and design approach. To get better use of this, please read [MMDetection Overview](https:\u002F\u002Fmmdetection.readthedocs.io\u002Fen\u002Flatest\u002Fget_started.html) for the first understanding of MMDetection.\n\nThe usage of MMYOLO is almost identical to MMDetection and all tutorials are straightforward to use, you can also learn about [MMDetection User Guide and Advanced Guide](https:\u002F\u002Fmmdetection.readthedocs.io\u002Fen\u002F3.x\u002F).\n\nFor different parts from MMDetection, we have also prepared user guides and advanced guides, please read our [documentation](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fzenh_CN\u002Flatest\u002F).\n\n\u003Cdetails>\n\u003Csummary>Get Started\u003C\u002Fsummary>\n\n- [Overview](docs\u002Fen\u002Fget_started\u002Foverview.md)\n- [Dependencies](docs\u002Fen\u002Fget_started\u002Fdependencies.md)\n- [Installation](docs\u002Fen\u002Fget_started\u002Finstallation.md)\n- [15 minutes object detection](docs\u002Fen\u002Fget_started\u002F15_minutes_object_detection.md)\n- [15 minutes rotated object detection](docs\u002Fen\u002Fget_started\u002F15_minutes_rotated_object_detection.md)\n- [15 minutes instance segmentation](docs\u002Fen\u002Fget_started\u002F15_minutes_instance_segmentation.md)\n- [Resources summary](docs\u002Fen\u002Fget_started\u002Farticle.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>Recommended Topics\u003C\u002Fsummary>\n\n- [How to contribute code to MMYOLO](docs\u002Fen\u002Frecommended_topics\u002Fcontributing.md)\n- [Training testing tricks](docs\u002Fen\u002Frecommended_topics\u002Ftraining_testing_tricks.md)\n- [MMYOLO model design](docs\u002Fen\u002Frecommended_topics\u002Fmodel_design.md)\n- [Algorithm principles and implementation](docs\u002Fen\u002Frecommended_topics\u002Falgorithm_descriptions\u002F)\n- [Replace the backbone network](docs\u002Fen\u002Frecommended_topics\u002Freplace_backbone.md)\n- [MMYOLO model complexity analysis](docs\u002Fen\u002Frecommended_topics\u002Fcomplexity_analysis.md)\n- [Annotation-to-deployment workflow for custom dataset](docs\u002Fen\u002Frecommended_topics\u002Flabeling_to_deployment_tutorials.md)\n- [Visualization](docs\u002Fen\u002Frecommended_topics\u002Fvisualization.md)\n- [Model deployment](docs\u002Fen\u002Frecommended_topics\u002Fdeploy\u002F)\n- [Troubleshooting steps](docs\u002Fen\u002Frecommended_topics\u002Ftroubleshooting_steps.md)\n- [MMYOLO application examples](docs\u002Fen\u002Frecommended_topics\u002Fapplication_examples\u002F)\n- [MM series repo essential basics](docs\u002Fen\u002Frecommended_topics\u002Fmm_basics.md)\n- [Dataset preparation and description](docs\u002Fen\u002Frecommended_topics\u002Fdataset_preparation.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>Common Usage\u003C\u002Fsummary>\n\n- [Resume training](docs\u002Fen\u002Fcommon_usage\u002Fresume_training.md)\n- [Enabling and disabling SyncBatchNorm](docs\u002Fen\u002Fcommon_usage\u002Fsyncbn.md)\n- [Enabling AMP](docs\u002Fen\u002Fcommon_usage\u002Famp_training.md)\n- [Multi-scale training and testing](docs\u002Fen\u002Fcommon_usage\u002Fms_training_testing.md)\n- [TTA Related Notes](docs\u002Fen\u002Fcommon_usage\u002Ftta.md)\n- [Add plugins to the backbone network](docs\u002Fen\u002Fcommon_usage\u002Fplugins.md)\n- [Freeze layers](docs\u002Fen\u002Fcommon_usage\u002Ffreeze_layers.md)\n- [Output model predictions](docs\u002Fen\u002Fcommon_usage\u002Foutput_predictions.md)\n- [Set random seed](docs\u002Fen\u002Fcommon_usage\u002Fset_random_seed.md)\n- [Module combination](docs\u002Fen\u002Fcommon_usage\u002Fmodule_combination.md)\n- [Cross-library calls using mim](docs\u002Fen\u002Fcommon_usage\u002Fmim_usage.md)\n- [Apply multiple Necks](docs\u002Fen\u002Fcommon_usage\u002Fmulti_necks.md)\n- [Specify specific device training or inference](docs\u002Fen\u002Fcommon_usage\u002Fspecify_device.md)\n- [Single and multi-channel application examples](docs\u002Fen\u002Fcommon_usage\u002Fsingle_multi_channel_applications.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>Useful Tools\u003C\u002Fsummary>\n\n- [Browse coco json](docs\u002Fen\u002Fuseful_tools\u002Fbrowse_coco_json.md)\n- [Browse dataset](docs\u002Fen\u002Fuseful_tools\u002Fbrowse_dataset.md)\n- [Print config](docs\u002Fen\u002Fuseful_tools\u002Fprint_config.md)\n- [Dataset analysis](docs\u002Fen\u002Fuseful_tools\u002Fdataset_analysis.md)\n- [Optimize anchors](docs\u002Fen\u002Fuseful_tools\u002Foptimize_anchors.md)\n- [Extract subcoco](docs\u002Fen\u002Fuseful_tools\u002Fextract_subcoco.md)\n- [Visualization scheduler](docs\u002Fen\u002Fuseful_tools\u002Fvis_scheduler.md)\n- [Dataset converters](docs\u002Fen\u002Fuseful_tools\u002Fdataset_converters.md)\n- [Download dataset](docs\u002Fen\u002Fuseful_tools\u002Fdownload_dataset.md)\n- [Log analysis](docs\u002Fen\u002Fuseful_tools\u002Flog_analysis.md)\n- [Model converters](docs\u002Fen\u002Fuseful_tools\u002Fmodel_converters.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>Basic Tutorials\u003C\u002Fsummary>\n\n- [Learn about configs with YOLOv5](docs\u002Fen\u002Ftutorials\u002Fconfig.md)\n- [Data flow](docs\u002Fen\u002Ftutorials\u002Fdata_flow.md)\n- [Rotated detection](docs\u002Fen\u002Ftutorials\u002Frotated_detection.md)\n- [Custom Installation](docs\u002Fen\u002Ftutorials\u002Fcustom_installation.md)\n- [Common Warning Notes](docs\u002Fzh_cn\u002Ftutorials\u002Fwarning_notes.md)\n- [FAQ](docs\u002Fen\u002Ftutorials\u002Ffaq.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>Advanced Tutorials\u003C\u002Fsummary>\n\n- [MMYOLO cross-library application](docs\u002Fen\u002Fadvanced_guides\u002Fcross-library_application.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>Descriptions\u003C\u002Fsummary>\n\n- [Changelog](docs\u002Fen\u002Fnotes\u002Fchangelog.md)\n- [Compatibility](docs\u002Fen\u002Fnotes\u002Fcompatibility.md)\n- [Conventions](docs\u002Fen\u002Fnotes\u002Fconventions.md)\n- [Code Style](docs\u002Fen\u002Fnotes\u002Fcode_style.md)\n\n\u003C\u002Fdetails>\n\n## 📊 Overview of Benchmark and Model Zoo [🔝](#-table-of-contents)\n\n\u003Cdiv align=center>\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_2886c95799d5.png\"\u002F>\n\u003C\u002Fdiv>\n\nResults and models are available in the [model zoo](docs\u002Fen\u002Fmodel_zoo.md).\n\n\u003Cdetails open>\n\u003Csummary>\u003Cb>Supported Tasks\u003C\u002Fb>\u003C\u002Fsummary>\n\n- [x] Object detection\n- [x] Rotated object detection\n\n\u003C\u002Fdetails>\n\n\u003Cdetails open>\n\u003Csummary>\u003Cb>Supported Algorithms\u003C\u002Fb>\u003C\u002Fsummary>\n\n- [x] [YOLOv5](configs\u002Fyolov5)\n- [ ] [YOLOv5u](configs\u002Fyolov5\u002Fyolov5u) (Inference only)\n- [x] [YOLOX](configs\u002Fyolox)\n- [x] [RTMDet](configs\u002Frtmdet)\n- [x] [RTMDet-Rotated](configs\u002Frtmdet)\n- [x] [YOLOv6](configs\u002Fyolov6)\n- [x] [YOLOv7](configs\u002Fyolov7)\n- [x] [PPYOLOE](configs\u002Fppyoloe)\n- [x] [YOLOv8](configs\u002Fyolov8)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails open>\n\u003Csummary>\u003Cb>Supported Datasets\u003C\u002Fb>\u003C\u002Fsummary>\n\n- [x] COCO Dataset\n- [x] VOC Dataset\n- [x] CrowdHuman Dataset\n- [x] DOTA 1.0 Dataset\n\n\u003C\u002Fdetails>\n\n\u003Cdetails open>\n\u003Cdiv align=\"center\">\n  \u003Cb>Module Components\u003C\u002Fb>\n\u003C\u002Fdiv>\n\u003Ctable align=\"center\">\n  \u003Ctbody>\n    \u003Ctr align=\"center\" valign=\"bottom\">\n      \u003Ctd>\n        \u003Cb>Backbones\u003C\u002Fb>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cb>Necks\u003C\u002Fb>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cb>Loss\u003C\u002Fb>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cb>Common\u003C\u002Fb>\n      \u003C\u002Ftd>\n    \u003C\u002Ftr>\n    \u003Ctr valign=\"top\">\n      \u003Ctd>\n      \u003Cul>\n        \u003Cli>YOLOv5CSPDarknet\u003C\u002Fli>\n        \u003Cli>YOLOv8CSPDarknet\u003C\u002Fli>\n        \u003Cli>YOLOXCSPDarknet\u003C\u002Fli>\n        \u003Cli>EfficientRep\u003C\u002Fli>\n        \u003Cli>CSPNeXt\u003C\u002Fli>\n        \u003Cli>YOLOv7Backbone\u003C\u002Fli>\n        \u003Cli>PPYOLOECSPResNet\u003C\u002Fli>\n        \u003Cli>mmdet backbone\u003C\u002Fli>\n        \u003Cli>mmcls backbone\u003C\u002Fli>\n        \u003Cli>timm\u003C\u002Fli>\n      \u003C\u002Ful>\n      \u003C\u002Ftd>\n      \u003Ctd>\n      \u003Cul>\n        \u003Cli>YOLOv5PAFPN\u003C\u002Fli>\n        \u003Cli>YOLOv8PAFPN\u003C\u002Fli>\n        \u003Cli>YOLOv6RepPAFPN\u003C\u002Fli>\n        \u003Cli>YOLOXPAFPN\u003C\u002Fli>\n        \u003Cli>CSPNeXtPAFPN\u003C\u002Fli>\n        \u003Cli>YOLOv7PAFPN\u003C\u002Fli>\n        \u003Cli>PPYOLOECSPPAFPN\u003C\u002Fli>\n      \u003C\u002Ful>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cul>\n          \u003Cli>IoULoss\u003C\u002Fli>\n          \u003Cli>mmdet loss\u003C\u002Fli>\n        \u003C\u002Ful>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cul>\n        \u003C\u002Ful>\n      \u003C\u002Ftd>\n    \u003C\u002Ftr>\n\u003C\u002Ftd>\n    \u003C\u002Ftr>\n  \u003C\u002Ftbody>\n\u003C\u002Ftable>\n\n\u003C\u002Fdetails>\n\n## ❓ FAQ [🔝](#-table-of-contents)\n\nPlease refer to the [FAQ](docs\u002Fen\u002Ftutorials\u002Ffaq.md) for frequently asked questions.\n\n## 🙌 Contributing [🔝](#-table-of-contents)\n\nWe appreciate all contributions to improving MMYOLO. Ongoing projects can be found in our [GitHub Projects](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fprojects). Welcome community users to participate in these projects. Please refer to [CONTRIBUTING.md](.github\u002FCONTRIBUTING.md) for the contributing guideline.\n\n## 🤝 Acknowledgement [🔝](#-table-of-contents)\n\nMMYOLO is an open source project that is contributed by researchers and engineers from various colleges and companies. We appreciate all the contributors who implement their methods or add new features, as well as users who give valuable feedback.\nWe wish that the toolbox and benchmark could serve the growing research community by providing a flexible toolkit to re-implement existing methods and develop their own new detectors.\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fgraphs\u002Fcontributors\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_3e9ae92f6879.png\"\u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>\n\n## 🖊️ Citation [🔝](#-table-of-contents)\n\nIf you find this project useful in your research, please consider citing:\n\n```latex\n@misc{mmyolo2022,\n    title={{MMYOLO: OpenMMLab YOLO} series toolbox and benchmark},\n    author={MMYOLO Contributors},\n    howpublished = {\\url{https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo}},\n    year={2022}\n}\n```\n\n## 🎫 License [🔝](#-table-of-contents)\n\nThis project is released under the [GPL 3.0 license](LICENSE).\n\n## 🏗️ Projects in OpenMMLab [🔝](#-table-of-contents)\n\n- [MMEngine](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmengine): OpenMMLab foundational library for training deep learning models.\n- [MMCV](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmcv): OpenMMLab foundational library for computer vision.\n- [MMPreTrain](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmpretrain): OpenMMLab pre-training toolbox and benchmark.\n- [MMagic](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmagic): Open**MM**Lab **A**dvanced, **G**enerative and **I**ntelligent **C**reation toolbox.\n- [MMDetection](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection): OpenMMLab detection toolbox and benchmark.\n- [MMDetection3D](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection3d): OpenMMLab's next-generation platform for general 3D object detection.\n- [MMRotate](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmrotate): OpenMMLab rotated object detection toolbox and benchmark.\n- [MMYOLO](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo): OpenMMLab YOLO series toolbox and benchmark.\n- [MMSegmentation](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmsegmentation): OpenMMLab semantic segmentation toolbox and benchmark.\n- [MMOCR](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmocr): OpenMMLab text detection, recognition, and understanding toolbox.\n- [MMPose](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmpose): OpenMMLab pose estimation toolbox and benchmark.\n- [MMHuman3D](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmhuman3d): OpenMMLab 3D human parametric model toolbox and benchmark.\n- [MMSelfSup](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmselfsup): OpenMMLab self-supervised learning toolbox and benchmark.\n- [MMRazor](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmrazor): OpenMMLab model compression toolbox and benchmark.\n- [MMFewShot](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmfewshot): OpenMMLab fewshot learning toolbox and benchmark.\n- [MMAction2](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmaction2): OpenMMLab's next-generation action understanding toolbox and benchmark.\n- [MMTracking](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmtracking): OpenMMLab video perception toolbox and benchmark.\n- [MMFlow](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmflow): OpenMMLab optical flow toolbox and benchmark.\n- [MMEditing](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmediting): OpenMMLab image and video editing toolbox.\n- [MMGeneration](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmgeneration): OpenMMLab image and video generative models toolbox.\n- [MMDeploy](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdeploy): OpenMMLab model deployment framework.\n- [MIM](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmim): MIM installs OpenMMLab packages.\n- [MMEval](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmeval): OpenMMLab machine learning evaluation library.\n- [Playground](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fplayground): A central hub for gathering and showcasing amazing projects built upon OpenMMLab.\n","\u003Cdiv align=\"center\">\n  \u003Cimg width=\"100%\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_28d28e0f87c2.png\"\u002F>\n  \u003Cdiv> \u003C\u002Fdiv>\n  \u003Cdiv align=\"center\">\n    \u003Cb>\u003Cfont size=\"5\">OpenMMLab 官网\u003C\u002Ffont>\u003C\u002Fb>\n    \u003Csup>\n      \u003Ca href=\"https:\u002F\u002Fopenmmlab.com\">\n        \u003Ci>\u003Cfont size=\"4\">HOT\u003C\u002Ffont>\u003C\u002Fi>\n      \u003C\u002Fa>\n    \u003C\u002Fsup>\n        \n    \u003Cb>\u003Cfont size=\"5\">OpenMMLab 平台\u003C\u002Ffont>\u003C\u002Fb>\n    \u003Csup>\n      \u003Ca href=\"https:\u002F\u002Fplatform.openmmlab.com\">\n        \u003Ci>\u003Cfont size=\"4\">立即体验\u003C\u002Ffont>\u003C\u002Fi>\n      \u003C\u002Fa>\n    \u003C\u002Fsup>\n  \u003C\u002Fdiv>\n  \u003Cdiv> \u003C\u002Fdiv>\n\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fmmyolo)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fmmyolo)\n[![docs](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdocs-latest-blue)](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002F)\n[![deploy](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fworkflows\u002Fdeploy\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Factions)\n[![codecov](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fopen-mmlab\u002Fmmyolo\u002Fbranch\u002Fmain\u002Fgraph\u002Fbadge.svg)](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fopen-mmlab\u002Fmmyolo)\n[![license](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fopen-mmlab\u002Fmmyolo.svg)](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002FLICENSE)\n[![open issues](https:\u002F\u002Fisitmaintained.com\u002Fbadge\u002Fopen\u002Fopen-mmlab\u002Fmmyolo.svg)](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues)\n[![issue resolution](https:\u002F\u002Fisitmaintained.com\u002Fbadge\u002Fresolution\u002Fopen-mmlab\u002Fmmyolo.svg)](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues)\n\n[📘文档](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002F) |\n[🛠️安装](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002Fget_started\u002Finstallation.html) |\n[👀模型 Zoo](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002Fmodel_zoo.html) |\n[🆕更新新闻](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002Fnotes\u002Fchangelog.html) |\n[🤔报告问题](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues\u002Fnew\u002Fchoose)\n\n\u003C\u002Fdiv>\n\n\u003Cdiv align=\"center\">\n\n英文 | [简体中文](README_zh-CN.md)\n\n\u003C\u002Fdiv>\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fopenmmlab.medium.com\u002F\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_062337b0e5ec.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Fdiscord.com\u002Fchannels\u002F1037617289144569886\u002F1046608014234370059\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_6342e5371027.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002FOpenMMLab\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_04c3beda0b07.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fopenmmlab\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_204fe79b5a90.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Fspace.bilibili.com\u002F1293512903\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_8655b6233577.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n  \u003Cimg src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F25839884\u002F218346358-56cc8e2f-a2b8-487f-9088-32480cceabcf.png\" width=\"3%\" alt=\"\" \u002F>\n  \u003Ca href=\"https:\u002F\u002Fwww.zhihu.com\u002Fpeople\u002Fopenmmlab\" style=\"text-decoration:none;\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_447c4737c11f.png\" width=\"3%\" alt=\"\" \u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>\n\n## 📄 目录\n\n- [🥳 🚀 新功能](#--whats-new-)\n  - [✨ 亮点](#-highlight-)\n- [📖 简介](#-introduction-)\n- [🛠️ 安装](#%EF%B8%8F-installation-)\n- [👨‍🏫 教程](#-tutorial-)\n- [📊 基准与模型 Zoo 概览](#-overview-of-benchmark-and-model-zoo-)\n- [❓ 常见问题](#-faq-)\n- [🙌 贡献](#-contributing-)\n- [🤝 致谢](#-acknowledgement-)\n- [🖊️ 引用](#️-citation-)\n- [🎫 许可证](#-license-)\n- [🏗️ OpenMMLab 中的项目](#%EF%B8%8F-projects-in-openmmlab-)\n\n## 🥳 🚀 新功能 [🔝](#-table-of-contents)\n\n💎 **v0.6.0** 于 2023年8月15日发布：\n\n- 支持 YOLOv5 实例分割\n- 支持基于 MMPose 的 YOLOX-Pose\n- 新增 15 分钟实例分割教程。\n- YOLOv5 支持使用掩码标注优化边界框\n- 新增多尺度训练与测试文档\n\n有关发布历史及更新详情，请参阅 [changelog](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest\u002Fnotes\u002Fchangelog.html)。\n\n### ✨ 重点 [🔝](#-table-of-contents)\n\n我们很高兴地宣布我们在实时目标识别任务上的最新成果——**RTMDet**，这是一系列全卷积单阶段检测器。RTMDet不仅在从小模型到超大模型的目标检测任务中实现了最佳的参数与精度权衡，还在实例分割和旋转目标检测任务上取得了新的SOTA性能。详细信息请参见[技术报告](https:\u002F\u002Farxiv.org\u002Fabs\u002F2212.07784)。预训练模型可在此处获取：[configs\u002Frtmdet](configs\u002Frtmdet)。\n\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frtmdet-an-empirical-study-of-designing-real\u002Freal-time-instance-segmentation-on-mscoco)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002Freal-time-instance-segmentation-on-mscoco?p=rtmdet-an-empirical-study-of-designing-real)\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frtmdet-an-empirical-study-of-designing-real\u002Fobject-detection-in-aerial-images-on-dota-1)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002Fobject-detection-in-aerial-images-on-dota-1?p=rtmdet-an-empirical-study-of-designing-real)\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frtmdet-an-empirical-study-of-designing-real\u002Fobject-detection-in-aerial-images-on-hrsc2016)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002Fobject-detection-in-aerial-images-on-hrsc2016?p=rtmdet-an-empirical-study-of-designing-real)\n\n| 任务                     | 数据集 | AP                                   | FPS(TRT FP16 BS1 3090) |\n| ------------------------ | ------- | ------------------------------------ | ---------------------- |\n| 目标检测         | COCO    | 52.8                                 | 322                    |\n| 实例分割    | COCO    | 44.6                                 | 188                    |\n| 旋转目标检测 | DOTA    | 78.9（单尺度）\u002F81.3（多尺度） | 121                    |\n\n\u003Cdiv align=center>\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_b5b09232c93f.png\"\u002F>\n\u003C\u002Fdiv>\n\nMMYOLO目前实现了目标检测和旋转目标检测算法，但与MMDetection版本相比，其训练速度显著提升。训练速度比之前版本快了2.6倍。\n\n## 📖 简介 [🔝](#-table-of-contents)\n\nMMYOLO是一个基于PyTorch和[MMDetection](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection)的YOLO系列算法开源工具箱。它是[OpenMMLab](https:\u002F\u002Fopenmmlab.com\u002F)项目的一部分。\n\n主分支支持**PyTorch 1.6及以上版本**。\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_6ddac5574998.gif\"\u002F>\n\n\u003Cdetails open>\n\u003Csummary>主要特性\u003C\u002Fsummary>\n\n- 🕹️ **统一且便捷的基准测试**\n\n  MMYOLO统一了各类YOLO算法中模块的实现，并提供了统一的基准测试。用户可以以公平、便捷的方式进行对比与分析。\n\n- 📚 **丰富而详尽的文档**\n\n  MMYOLO提供了丰富的入门指南、模型部署、高级用法以及算法分析文档，使不同水平的用户都能轻松上手并快速扩展功能。\n\n- 🧩 **模块化设计**\n\n  MMYOLO将框架分解为不同的组件，用户可以通过组合不同模块并搭配多种训练与测试策略，轻松自定义模型。\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_c5633eb648bb.jpg\" alt=\"BaseModule-P5\"\u002F>\n  上图由RangeKing@GitHub贡献，非常感谢！\n\n而P6模型的示意图则位于[model_design.md](docs\u002Fen\u002Frecommended_topics\u002Fmodel_design.md)。\n\n\u003C\u002Fdetails>\n\n## 🛠️ 安装 [🔝](#-table-of-contents)\n\nMMYOLO依赖于PyTorch、MMCV、MMEngine和MMDetection。以下是快速安装步骤。更多详细说明请参阅[安装指南](docs\u002Fen\u002Fget_started\u002Finstallation.md)。\n\n```shell\nconda create -n mmyolo python=3.8 pytorch==1.10.1 torchvision==0.11.2 cudatoolkit=11.3 -c pytorch -y\nconda activate mmyolo\npip install openmim\nmim install \"mmengine>=0.6.0\"\nmim install \"mmcv>=2.0.0rc4,\u003C2.1.0\"\nmim install \"mmdet>=3.0.0,\u003C4.0.0\"\ngit clone https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo.git\ncd mmyolo\n# 安装albumentations\npip install -r requirements\u002Falbu.txt\n# 安装MMYOLO\nmim install -v -e .\n```\n\n## 👨‍🏫 教程 [🔝](#-table-of-contents)\n\nMMYOLO 基于 MMDetection 构建，采用相同的代码结构和设计思路。为更好地使用本项目，请先阅读 [MMDetection 概述](https:\u002F\u002Fmmdetection.readthedocs.io\u002Fen\u002Flatest\u002Fget_started.html)，以对 MMDetection 有一个初步了解。\n\nMMYOLO 的使用方法与 MMDetection 几乎完全一致，所有教程都简单易懂；您也可以参考 [MMDetection 用户指南与进阶指南](https:\u002F\u002Fmmdetection.readthedocs.io\u002Fen\u002F3.x\u002F) 进行学习。\n\n针对 MMDetection 中的各个不同模块，我们也准备了相应的用户指南与进阶指南，请参阅我们的 [文档](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fzenh_CN\u002Flatest\u002F)。\n\n\u003Cdetails>\n\u003Csummary>入门指南\u003C\u002Fsummary>\n\n- [概述](docs\u002Fen\u002Fget_started\u002Foverview.md)\n- [依赖项](docs\u002Fen\u002Fget_started\u002Fdependencies.md)\n- [安装](docs\u002Fen\u002Fget_started\u002Finstallation.md)\n- [15 分钟完成目标检测](docs\u002Fen\u002Fget_started\u002F15_minutes_object_detection.md)\n- [15 分钟完成旋转目标检测](docs\u002Fen\u002Fget_started\u002F15_minutes_rotated_object_detection.md)\n- [15 分钟完成实例分割](docs\u002Fen\u002Fget_started\u002F15_minutes_instance_segmentation.md)\n- [资源汇总](docs\u002Fen\u002Fget_started\u002Farticle.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>推荐主题\u003C\u002Fsummary>\n\n- [如何向 MMYOLO 贡献代码](docs\u002Fen\u002Frecommended_topics\u002Fcontributing.md)\n- [训练与测试技巧](docs\u002Fen\u002Frecommended_topics\u002Ftraining_testing_tricks.md)\n- [MMYOLO 模型设计](docs\u002Fen\u002Frecommended_topics\u002Fmodel_design.md)\n- [算法原理与实现](docs\u002Fen\u002Frecommended_topics\u002Falgorithm_descriptions\u002F)\n- [更换主干网络](docs\u002Fen\u002Frecommended_topics\u002Freplace_backbone.md)\n- [MMYOLO 模型复杂度分析](docs\u002Fen\u002Frecommended_topics\u002Fcomplexity_analysis.md)\n- [自定义数据集的标注到部署流程](docs\u002Fen\u002Frecommended_topics\u002Flabeling_to_deployment_tutorials.md)\n- [可视化](docs\u002Fen\u002Frecommended_topics\u002Fvisualization.md)\n- [模型部署](docs\u002Fen\u002Frecommended_topics\u002Fdeploy\u002F)\n- [故障排查步骤](docs\u002Fen\u002Frecommended_topics\u002Ftroubleshooting_steps.md)\n- [MMYOLO 应用示例](docs\u002Fen\u002Frecommended_topics\u002Fapplication_examples\u002F)\n- [MM 系列仓库基础要点](docs\u002Fen\u002Frecommended_topics\u002Fmm_basics.md)\n- [数据集准备与说明](docs\u002Fen\u002Frecommended_topics\u002Fdataset_preparation.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>常用操作\u003C\u002Fsummary>\n\n- [恢复训练](docs\u002Fen\u002Fcommon_usage\u002Fresume_training.md)\n- [启用与禁用 SyncBatchNorm](docs\u002Fen\u002Fcommon_usage\u002Fsyncbn.md)\n- [启用 AMP](docs\u002Fen\u002Fcommon_usage\u002Famp_training.md)\n- [多尺度训练与测试](docs\u002Fen\u002Fcommon_usage\u002Fms_training_testing.md)\n- [TTA 相关说明](docs\u002Fen\u002Fcommon_usage\u002Ftta.md)\n- [为主干网络添加插件](docs\u002Fen\u002Fcommon_usage\u002Fplugins.md)\n- [冻结层](docs\u002Fen\u002Fcommon_usage\u002Ffreeze_layers.md)\n- [输出模型预测结果](docs\u002Fen\u002Fcommon_usage\u002Foutput_predictions.md)\n- [设置随机种子](docs\u002Fen\u002Fcommon_usage\u002Fset_random_seed.md)\n- [模块组合](docs\u002Fen\u002Fcommon_usage\u002Fmodule_combination.md)\n- [使用 mim 进行跨库调用](docs\u002Fen\u002Fcommon_usage\u002Fmim_usage.md)\n- [应用多个 Neck](docs\u002Fen\u002Fcommon_usage\u002Fmulti_necks.md)\n- [指定特定设备进行训练或推理](docs\u002Fen\u002Fcommon_usage\u002Fspecify_device.md)\n- [单通道与多通道应用示例](docs\u002Fen\u002Fcommon_usage\u002Fsingle_multi_channel_applications.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>实用工具\u003C\u002Fsummary>\n\n- [浏览 COCO JSON 文件](docs\u002Fen\u002Fuseful_tools\u002Fbrowse_coco_json.md)\n- [浏览数据集](docs\u002Fen\u002Fuseful_tools\u002Fbrowse_dataset.md)\n- [打印配置文件](docs\u002Fen\u002Fuseful_tools\u002Fprint_config.md)\n- [数据集分析](docs\u002Fen\u002Fuseful_tools\u002Fdataset_analysis.md)\n- [优化锚点](docs\u002Fen\u002Fuseful_tools\u002Foptimize_anchors.md)\n- [提取子 COCO 数据集](docs\u002Fen\u002Fuseful_tools\u002Fextract_subcoco.md)\n- [可视化调度器](docs\u002Fen\u002Fuseful_tools\u002Fvis_scheduler.md)\n- [数据集转换工具](docs\u002Fen\u002Fuseful_tools\u002Fdataset_converters.md)\n- [下载数据集](docs\u002Fen\u002Fuseful_tools\u002Fdownload_dataset.md)\n- [日志分析](docs\u002Fen\u002Fuseful_tools\u002Flog_analysis.md)\n- [模型转换工具](docs\u002Fen\u002Fuseful_tools\u002Fmodel_converters.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>基础教程\u003C\u002Fsummary>\n\n- [通过 YOLOv5 学习配置文件](docs\u002Fen\u002Ftutorials\u002Fconfig.md)\n- [数据流](docs\u002Fen\u002Ftutorials\u002Fdata_flow.md)\n- [旋转目标检测](docs\u002Fen\u002Ftutorials\u002Frotated_detection.md)\n- [自定义安装](docs\u002Fen\u002Ftutorials\u002Fcustom_installation.md)\n- [常见警告说明](docs\u002Fzh_cn\u002Ftutorials\u002Fwarning_notes.md)\n- [常见问题解答](docs\u002Fen\u002Ftutorials\u002Ffaq.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>进阶教程\u003C\u002Fsummary>\n\n- [MMYOLO 跨库应用](docs\u002Fen\u002Fadvanced_guides\u002Fcross-library_application.md)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>说明\u003C\u002Fsummary>\n\n- [更新日志](docs\u002Fen\u002Fnotes\u002Fchangelog.md)\n- [兼容性](docs\u002Fen\u002Fnotes\u002Fcompatibility.md)\n- [约定](docs\u002Fen\u002Fnotes\u002Fconventions.md)\n- [代码风格](docs\u002Fen\u002Fnotes\u002Fcode_style.md)\n\n\u003C\u002Fdetails>\n\n## 📊 基准与模型库概览 [🔝](#-table-of-contents)\n\n\u003Cdiv align=center>\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_2886c95799d5.png\"\u002F>\n\u003C\u002Fdiv>\n\n结果与模型可在[模型库](docs\u002Fen\u002Fmodel_zoo.md)中获取。\n\n\u003Cdetails open>\n\u003Csummary>\u003Cb>支持的任务\u003C\u002Fb>\u003C\u002Fsummary>\n\n- [x] 目标检测\n- [x] 旋转目标检测\n\n\u003C\u002Fdetails>\n\n\u003Cdetails open>\n\u003Csummary>\u003Cb>支持的算法\u003C\u002Fb>\u003C\u002Fsummary>\n\n- [x] [YOLOv5](configs\u002Fyolov5)\n- [ ] [YOLOv5u](configs\u002Fyolov5\u002Fyolov5u)（仅用于推理）\n- [x] [YOLOX](configs\u002Fyolox)\n- [x] [RTMDet](configs\u002Frtmdet)\n- [x] [RTMDet-Rotated](configs\u002Frtmdet)\n- [x] [YOLOv6](configs\u002Fyolov6)\n- [x] [YOLOv7](configs\u002Fyolov7)\n- [x] [PPYOLOE](configs\u002Fppyoloe)\n- [x] [YOLOv8](configs\u002Fyolov8)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails open>\n\u003Csummary>\u003Cb>支持的数据集\u003C\u002Fb>\u003C\u002Fsummary>\n\n- [x] COCO 数据集\n- [x] VOC 数据集\n- [x] CrowdHuman 数据集\n- [x] DOTA 1.0 数据集\n\n\u003C\u002Fdetails>\n\n\u003Cdetails open>\n\u003Cdiv align=\"center\">\n  \u003Cb>模块组件\u003C\u002Fb>\n\u003C\u002Fdiv>\n\u003Ctable align=\"center\">\n  \u003Ctbody>\n    \u003Ctr align=\"center\" valign=\"bottom\">\n      \u003Ctd>\n        \u003Cb>骨干网络\u003C\u002Fb>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cb>颈部网络\u003C\u002Fb>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cb>损失函数\u003C\u002Fb>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cb>通用模块\u003C\u002Fb>\n      \u003C\u002Ftd>\n    \u003C\u002Ftr>\n    \u003Ctr valign=\"top\">\n      \u003Ctd>\n      \u003Cul>\n        \u003Cli>YOLOv5CSPDarknet\u003C\u002Fli>\n        \u003Cli>YOLOv8CSPDarknet\u003C\u002Fli>\n        \u003Cli>YOLOXCSPDarknet\u003C\u002Fli>\n        \u003Cli>EfficientRep\u003C\u002Fli>\n        \u003Cli>CSPNeXt\u003C\u002Fli>\n        \u003Cli>YOLOv7Backbone\u003C\u002Fli>\n        \u003Cli>PPYOLOECSPResNet\u003C\u002Fli>\n        \u003Cli>mmdet 骨干网络\u003C\u002Fli>\n        \u003Cli>mmcls 骨干网络\u003C\u002Fli>\n        \u003Cli>timm\u003C\u002Fli>\n      \u003C\u002Ful>\n      \u003C\u002Ftd>\n      \u003Ctd>\n      \u003Cul>\n        \u003Cli>YOLOv5PAFPN\u003C\u002Fli>\n        \u003Cli>YOLOv8PAFPN\u003C\u002Fli>\n        \u003Cli>YOLOv6RepPAFPN\u003C\u002Fli>\n        \u003Cli>YOLOXPAFPN\u003C\u002Fli>\n        \u003Cli>CSPNeXtPAFPN\u003C\u002Fli>\n        \u003Cli>YOLOv7PAFPN\u003C\u002Fli>\n        \u003Cli>PPYOLOECSPPAFPN\u003C\u002Fli>\n      \u003C\u002Ful>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cul>\n          \u003Cli>IoULoss\u003C\u002Fli>\n          \u003Cli>mmdet 损失函数\u003C\u002Fli>\n        \u003C\u002Ful>\n      \u003C\u002Ftd>\n      \u003Ctd>\n        \u003Cul>\n        \u003C\u002Ful>\n      \u003C\u002Ftd>\n    \u003C\u002Ftr>\n\u003C\u002Ftd>\n    \u003C\u002Ftr>\n  \u003C\u002Ftbody>\n\u003C\u002Ftable>\n\n\u003C\u002Fdetails>\n\n## ❓ 常见问题 [🔝](#-table-of-contents)\n\n请参阅[常见问题](docs\u002Fen\u002Ftutorials\u002Ffaq.md)以获取常见问题解答。\n\n## 🙌 贡献 [🔝](#-table-of-contents)\n\n我们非常感谢所有为改进 MMYOLO 所做的贡献。正在进行的项目可在我们的[GitHub 项目](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fprojects)中找到。欢迎社区用户参与这些项目。有关贡献指南，请参阅[CONTRIBUTING.md](.github\u002FCONTRIBUTING.md)。\n\n## 🤝 致谢 [🔝](#-table-of-contents)\n\nMMYOLO 是一个由来自不同院校和公司的研究人员与工程师共同贡献的开源项目。我们感谢所有实现其方法或添加新功能的贡献者，以及提供宝贵反馈的用户。\n我们希望该工具箱与基准能够服务于不断壮大的研究社区，为其提供灵活的工具包，以便重新实现现有方法并开发自己的新型检测器。\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fgraphs\u002Fcontributors\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_readme_3e9ae92f6879.png\"\u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>\n\n## 🖊️ 引用 [🔝](#-table-of-contents)\n\n如果您在研究中发现本项目有用，请考虑引用：\n\n```latex\n@misc{mmyolo2022,\n    title={{MMYOLO: OpenMMLab YOLO} 系列工具箱与基准},\n    author={MMYOLO 贡献者},\n    howpublished = {\\url{https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo}},\n    year={2022}\n}\n```\n\n## 🎫 许可证 [🔝](#-table-of-contents)\n\n本项目根据[GPL 3.0 许可证](LICENSE)发布。\n\n## 🏗️ OpenMMLab 中的项目 [🔝](#-table-of-contents)\n\n- [MMEngine](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmengine): OpenMMLab 用于训练深度学习模型的基础库。\n- [MMCV](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmcv): OpenMMLab 用于计算机视觉的基础库。\n- [MMPreTrain](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmpretrain): OpenMMLab 预训练工具箱与基准。\n- [MMagic](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmagic): Open**MM**Lab **A**dvanced, **G**enerative and **I**ntelligent **C**reation 工具箱。\n- [MMDetection](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection): OpenMMLab 检测工具箱与基准。\n- [MMDetection3D](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection3d): OpenMMLab 下一代通用 3D 目标检测平台。\n- [MMRotate](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmrotate): OpenMMLab 旋转目标检测工具箱与基准。\n- [MMYOLO](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo): OpenMMLab YOLO 系列工具箱与基准。\n- [MMSegmentation](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmsegmentation): OpenMMLab 语义分割工具箱与基准。\n- [MMOCR](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmocr): OpenMMLab 文本检测、识别与理解工具箱。\n- [MMPose](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmpose): OpenMMLab 姿势估计工具箱与基准。\n- [MMHuman3D](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmhuman3d): OpenMMLab 3D 人体参数化模型工具箱与基准。\n- [MMSelfSup](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmselfsup): OpenMMLab 自监督学习工具箱与基准。\n- [MMRazor](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmrazor): OpenMMLab 模型压缩工具箱与基准。\n- [MMFewShot](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmfewshot): OpenMMLab 少样本学习工具箱与基准。\n- [MMAction2](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmaction2): OpenMMLab 下一代动作理解工具箱与基准。\n- [MMTracking](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmtracking): OpenMMLab 视频感知工具箱与基准。\n- [MMFlow](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmflow): OpenMMLab 光流工具箱与基准。\n- [MMEditing](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmediting): OpenMMLab 图像与视频编辑工具箱。\n- [MMGeneration](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmgeneration): OpenMMLab 图像与视频生成模型工具箱。\n- [MMDeploy](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdeploy): OpenMMLab 模型部署框架。\n- [MIM](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmim): MIM 安装 OpenMMLab 包。\n- [MMEval](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmeval): OpenMMLab 机器学习评估库。\n- [Playground](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fplayground): 一个汇集并展示基于 OpenMMLab 构建的精彩项目的中心枢纽。","# MMYOLO 快速上手指南\n\nMMYOLO 是基于 PyTorch 和 MMDetection 的开源工具箱，专注于 YOLO 系列算法（包括目标检测、旋转目标检测和实例分割）。它提供了统一的基准测试、模块化设计以及丰富的文档，旨在帮助用户快速复现和研究最新的 YOLO 算法。\n\n## 1. 环境准备\n\n在开始之前，请确保您的系统满足以下要求：\n\n*   **操作系统**: Linux (推荐), Windows, macOS\n*   **Python**: 3.8 及以上版本\n*   **PyTorch**: 1.6+ (推荐 1.10.1+)\n*   **CUDA**: 根据显卡驱动版本安装对应的 CUDA Toolkit (示例中使用 11.3)\n*   **前置依赖**: MMYOLO 依赖 `MMEngine`, `MMCV`, 和 `MMDetection`。我们将通过 `OpenMIM` 工具自动管理这些依赖。\n\n> **提示**：建议使用 Conda 创建独立的虚拟环境以避免依赖冲突。国内用户可配置清华源或阿里源加速包下载。\n\n## 2. 安装步骤\n\n请按照以下步骤在终端中执行命令进行安装：\n\n### 第一步：创建并激活 Conda 环境\n```shell\nconda create -n mmyolo python=3.8 pytorch==1.10.1 torchvision==0.11.2 cudatoolkit=11.3 -c pytorch -y\nconda activate mmyolo\n```\n*(注：国内用户若下载缓慢，可在命令后添加 `-c https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fanaconda\u002Fpkgs\u002Fmain\u002F` 等镜像源参数，或在 `.condarc` 中全局配置)*\n\n### 第二步：安装 OpenMIM 及核心依赖\n使用 `mim` 工具安装 MM 系列基础库，它会自动处理版本兼容性：\n```shell\npip install openmim\nmim install \"mmengine>=0.6.0\"\nmim install \"mmcv>=2.0.0rc4,\u003C2.1.0\"\nmim install \"mmdet>=3.0.0,\u003C4.0.0\"\n```\n\n### 第三步：克隆代码并安装 MMYOLO\n```shell\ngit clone https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo.git\ncd mmyolo\n\n# 安装 albumentations 数据增强库\npip install -r requirements\u002Falbu.txt\n\n# 以可编辑模式安装 MMYOLO\nmim install -v -e .\n```\n\n## 3. 基本使用\n\nMMYOLO 的使用流程与 MMDetection 高度一致，主要包括**数据准备**、**模型训练**、**模型测试**和**可视化**。\n\n### 3.1 准备数据集\nMMYOLO 默认支持 COCO 格式数据集。您需要将数据集整理为如下结构，并修改配置文件中的 `data_root` 路径：\n```text\ndata\n├── coco\n│   ├── annotations\n│   │   ├── instances_train2017.json\n│   │   └── instances_val2017.json\n│   ├── train2017\n│   └── val2017\n```\n\n### 3.2 快速训练 (15 分钟入门)\nMMYOLO 提供了预配置的配置文件。以下以训练 YOLOv5 为例，使用单张 GPU 进行训练：\n\n```shell\n# 使用默认的 YOLOv5 配置文件启动训练\n# --work-dir 指定工作目录，--auto-scale-lr 自动调整学习率\npython tools\u002Ftrain.py configs\u002Fyolov5\u002Fyolov5_s-v61_syncbn_fast_8xb16-300e_coco.py\n```\n\n若要使用多卡训练（例如 8 张卡）：\n```shell\nbash tools\u002Fdist_train.sh configs\u002Fyolov5\u002Fyolov5_s-v61_syncbn_fast_8xb16-300e_coco.py 8\n```\n\n### 3.3 模型测试与评估\n训练完成后，使用生成的权重文件进行测试：\n\n```shell\n# 单卡测试\npython tools\u002Ftest.py configs\u002Fyolov5\u002Fyolov5_s-v61_syncbn_fast_8xb16-300e_coco.py work_dirs\u002Fyolov5_s-v61_syncbn_fast_8xb16-300e_coco\u002Fepoch_300.pth\n\n# 多卡测试\nbash tools\u002Fdist_test.sh configs\u002Fyolov5\u002Fyolov5_s-v61_syncbn_fast_8xb16-300e_coco.py work_dirs\u002Fyolov5_s-v61_syncbn_fast_8xb16-300e_coco\u002Fepoch_300.pth 8\n```\n\n### 3.4 推理与可视化\n对单张图片进行推理并保存结果：\n\n```shell\npython tools\u002Fdemo\u002Fimage_demo.py demo\u002Fdemo.jpg configs\u002Fyolov5\u002Fyolov5_s-v61_syncbn_fast_8xb16-300e_coco.py work_dirs\u002Fyolov5_s-v61_syncbn_fast_8xb16-300e_coco\u002Fepoch_300.pth --out-file result.jpg\n```\n\n### 3.5 进阶任务\nMMYOLO 同样支持旋转目标检测和实例分割，只需更换对应的配置文件即可：\n*   **旋转目标检测**: 使用 `configs\u002Frotated_yolov5\u002F` 下的配置。\n*   **实例分割**: 使用 `configs\u002Fyolov5_inst_seg\u002F` 下的配置。\n\n更多详细教程（如自定义数据集、模型部署、算法原理分析）请访问 [MMYOLO 官方文档](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fzh_CN\u002Flatest\u002F)。","某智慧物流团队正致力于升级其自动化分拣系统，需要快速部署高精度模型以识别传送带上不同朝向的包裹和条码。\n\n### 没有 mmyolo 时\n- **算法选型困难**：面对 YOLOv5 到 YOLOv8 以及 YOLOX 等众多版本，团队需分别查找不同作者的代码库，环境依赖冲突频发，难以统一评估性能。\n- **旋转目标支持缺失**：传统检测器难以处理随意堆叠导致的倾斜包裹，团队需从零复现复杂的旋转检测算法（如 RTMDet-Rotated），研发周期长达数周。\n- **训练调优成本高**：缺乏统一的基准测试（Benchmark）和数据增强策略，每次调整超参数都需手动修改底层代码，试错效率极低。\n- **部署落地繁琐**：不同模型架构的输出格式各异，导出为 TensorRT 或 ONNX 时需编写大量适配脚本，工程化落地阻力大。\n\n### 使用 mmyolo 后\n- **一站式模型库**：mmyolo 集成了主流 YOLO 系列及 RTMDet 等先进算法，提供统一的接口和预训练权重，团队一天内即可完成多模型对比验证。\n- **原生旋转检测**：直接调用内置的 RTMDet-Rotated 模块，无需额外开发即可精准识别任意角度包裹，显著提升了复杂堆叠场景下的检出率。\n- **标准化训练流程**：依托模块化设计，团队成员仅需修改配置文件即可切换主干网络或调整增强策略，实验迭代速度提升 3 倍以上。\n- **无缝部署体验**：工具链原生支持多种后端部署格式，一键导出优化模型，将算法从训练到产线部署的时间从一周缩短至两天。\n\nmmyolo 通过统一化的架构设计和丰富的算法储备，将物流场景下的目标检测研发从“重复造轮子”转变为高效的“积木式创新”。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-mmlab_mmyolo_28d28e0f.png","open-mmlab","OpenMMLab","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fopen-mmlab_7c171dd7.png","",null,"https:\u002F\u002Fopenmmlab.com","https:\u002F\u002Fgithub.com\u002Fopen-mmlab",[83,87,91,95,98],{"name":84,"color":85,"percentage":86},"Python","#3572A5",99.5,{"name":88,"color":89,"percentage":90},"C++","#f34b7d",0.2,{"name":92,"color":93,"percentage":94},"Shell","#89e051",0.1,{"name":96,"color":97,"percentage":94},"Dockerfile","#384d54",{"name":99,"color":100,"percentage":94},"CMake","#DA3434",3429,627,"2026-04-05T02:38:11","GPL-3.0","未说明","需要 NVIDIA GPU (基于 CUDA)，示例中使用了 RTX 3090，CUDA 11.3+",{"notes":108,"python":109,"dependencies":110},"主分支支持 PyTorch 1.6+。官方安装示例使用 Python 3.8、PyTorch 1.10.1 和 CUDA 11.3。建议使用 conda 创建虚拟环境并通过 OpenMIM 工具安装依赖。该工具是基于 MMDetection 的 YOLO 系列算法工具箱。","3.8+",[111,112,113,114,115,116],"torch>=1.6","torchvision","mmengine>=0.6.0","mmcv>=2.0.0rc4,\u003C2.1.0","mmdet>=3.0.0,\u003C4.0.0","albumentations",[14,13],[119,120,121,122,123,124,125,126,127,128,129,130],"object-detection","pytorch","yolo","yolov5","yolov6","yolox","rtmdet","yolov7","yolov8","ppyoloe","deep-learning","rotated-object-detection","2026-03-27T02:49:30.150509","2026-04-06T07:12:38.514909",[134,139,144,149,154,158],{"id":135,"question_zh":136,"answer_zh":137,"source_url":138},5951,"使用自定义 COCO 数据集训练 YOLOX 时，cls 和 reg 的 loss 始终为 0，如何解决？","这是因为在配置文件的 `train_dataloader` 中未设置 `metainfo`。需要在数据加载器的配置中显式添加类别信息。例如：\n```python\nmetainfo = {\n    'classes': ('your_class1', 'your_class2'),\n    'palette': [(220, 20, 60), (119, 11, 32)]\n}\n```\n确保 `train_dataloader['dataset']` 中包含此 `metainfo` 字段，否则模型无法正确映射类别标签，导致损失为 0。","https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues\u002F405",{"id":140,"question_zh":141,"answer_zh":142,"source_url":143},5952,"训练自定义数据集（如气球检测）时遇到 `IndexError: list index out of range` 错误，提示 `data['category_id'] = self.cat_ids[label]`，怎么办？","该错误通常是因为未在验证集数据加载器 `val_dataloader['dataset']` 中配置 `metainfo`。请检查配置文件，确保验证集也添加了类别元信息：\n```python\nmetainfo = {\n    'classes': ('balloon', ),\n    'palette': [(220, 20, 60),]\n}\n```\n注意：MMYOLO 0.3.0 及以上版本要求类别名称使用小写字母。此外，如果是自定义数据集，建议先生成针对该数据集的 anchors 再进行训练。","https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues\u002F420",{"id":145,"question_zh":146,"answer_zh":147,"source_url":148},5953,"如何在 MMYOLO 中推理单张图片？","虽然 Issue 标题询问如何推理，但具体操作通常通过 `demo\u002Fimage_demo.py` 脚本执行。命令格式如下：\n```bash\npython demo\u002Fimage_demo.py \u003Cimage_path> \u003Cconfig_file> \u003Ccheckpoint_file> --device cuda:0\n```\n如果输出结果中的类别显示为数字而非名称，或者文字模糊，可以通过调整可视化代码中的 `putText` 参数（如字体大小、线条粗细）来优化显示效果。框架支持将类别 ID 自动映射为配置文件中定义的类别名称。","https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues\u002F958",{"id":150,"question_zh":151,"answer_zh":152,"source_url":153},5954,"训练 YOLOv7 自定义数据集时 loss 全为 0 且测试结果为空，可能是什么原因？","这种情况通常与数据标注或配置有关。首先检查是否像其他 YOLO 系列模型一样，忘记在 `train_dataloader` 和 `val_dataloader` 中设置 `metainfo`（包含 classes 和 palette）。其次，确认数据集的标注格式是否正确转换为 MMYOLO 支持的格式（如 COCO 格式），并且类别索引从 0 开始连续。如果使用了预生成的 anchors，请确保它们是基於当前自定义数据集重新计算生成的，而不是直接沿用官方默认值。","https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues\u002F447",{"id":155,"question_zh":156,"answer_zh":157,"source_url":148},5955,"推理结果图片上的类别标签显示为数字（cls），如何改为显示具体的类别名称？","这是可以实现的。推理脚本默认会使用配置文件中 `metainfo` 里定义的 `classes` 列表将类别索引映射为名称。如果仍然显示数字，请检查：\n1. 配置文件是否正确加载了包含 `classes` 字段的 `metainfo`。\n2. 可视化函数是否正确传递了类别名称列表。\n通常不需要修改核心代码，只需确保配置文件中的元信息正确即可，因为可视化逻辑（如 `putText`）会自动处理名称映射。",{"id":159,"question_zh":160,"answer_zh":161,"source_url":162},5956,"SAHI（切片辅助推理）功能在 MMYOLO 中是如何规划的？是否支持在线训练？","SAHI 主要用于大图像（如卫星图像）的切片推理。目前的规划是将其作为一个工具（tool）集成，以便能应用于多种目标检测和实例分割模型，而不是作为单独的模型。关于是否支持在线训练（online training），社区曾有询问，但该功能主要侧重于推理阶段的切片处理。用户可以尝试将其作为推理工具使用，具体的集成进度和示例代码需关注官方后续的更新或迁移到 MMDetection 后的文档。","https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fissues\u002F230",[164,169,174,179,184,189,194,199,204],{"id":165,"version":166,"summary_zh":167,"released_at":168},105565,"v0.6.0","## v0.6.0 (15\u002F8\u002F2023)\r\n\r\n### Highlights\r\n\r\n- Support YOLOv5 instance segmentation\r\n- Support YOLOX-Pose based on MMPose\r\n- Add 15 minutes instance segmentation tutorial.\r\n- YOLOv5 supports using mask annotation to optimize bbox\r\n- Add Multi-scale training and testing docs\r\n\r\n### New Features\r\n\r\n- Add training and testing tricks doc (#659)\r\n- Support setting the cache_size_limit parameter and support mmdet 3.0.0 (#707)\r\n- Support YOLOv5u and YOLOv6 3.0 inference (#624, #744)\r\n- Support model-only inference (#733)\r\n- Add YOLOv8 deepstream config (#633)\r\n- Add ionogram example in MMYOLO application (#643)\r\n\r\n### Bug Fixes\r\n\r\n- Fix the browse_dataset for visualization of test and val (#641)\r\n- Fix installation doc error (#662)\r\n- Fix yolox-l ckpt link (#677)\r\n- Fix typos in the YOLOv7 and YOLOv8 diagram (#621, #710)\r\n- Adjust the order of package imports in `boxam_vis_demo.py` (#655)\r\n\r\n### Improvements\r\n\r\n- Optimize the `convert_kd_ckpt_to_student.py` file (#647)\r\n- Add en doc of `FAQ` and `training_testing_tricks` (#691,#693)\r\n\r\n### Contributors\r\n\r\nA total of 21 developers contributed to this release.\r\n\r\nThank @Lum1104,@azure-wings,@FeiGeChuanShu,@Lingrui Gu,@Nioolek,@huayuan4396,@RangeKing,@danielhonies,@yechenzhi,@JosonChan1998,@kitecats,@Qingrenn,@triple-Mu,@kikefdezl,@zhangrui-wolf,@xin-li-67,@Ben-Louis,@zgzhengSEU,@VoyagerXvoyagerx,@tang576225574,@hhaAndroid\r\n","2023-08-15T05:20:22",{"id":170,"version":171,"summary_zh":172,"released_at":173},105566,"v0.5.0","\u003Cimg width=\"100%\" src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F27466624\u002F222385101-516e551c-49f5-480d-a135-4b24ee6dc308.png\"\u002F>\r\n\r\n\r\n## v0.5.0 (2\u002F3\u002F2023)\r\n\r\n### Highlights\r\n\r\n1. Support [RTMDet-R](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Frtmdet\u002FREADME.md#rotated-object-detection) rotated object detection\r\n2. Support for using mask annotation to improve [YOLOv8](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Fyolov8\u002FREADME.md) object detection performance\r\n3. Support [MMRazor](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Frazor\u002Fsubnets\u002FREADME.md) searchable NAS sub-network as the backbone of YOLO series algorithm\r\n4. Support calling [MMRazor](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Frtmdet\u002Fdistillation\u002FREADME.md) to distill the knowledge of RTMDet\r\n5. [MMYOLO](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fzh_CN\u002Fdev\u002F) document structure optimization, comprehensive content upgrade\r\n6. Improve YOLOX mAP and training speed based on RTMDet training hyperparameters\r\n7. Support calculation of model parameters and FLOPs, provide GPU latency data on T4 devices, and update [Model Zoo](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fdocs\u002Fen\u002Fmodel_zoo.md)\r\n8. Support test-time augmentation (TTA)\r\n9. Support RTMDet, YOLOv8 and YOLOv7 assigner visualization\r\n\r\n### New Features\r\n\r\n01. Support inference for RTMDet instance segmentation tasks (#583)\r\n02. Beautify the configuration file in MMYOLO and add more comments (#501, #506, #516, #529, #531, #539)\r\n03. Refactor and optimize documentation (#568, #573, #579, #584, #587, #589, #596, #599, #600)\r\n04. Support fast version of YOLOX (#518)\r\n05. Support DeepStream in EasyDeploy and add documentation (#485, #545, #571)\r\n06. Add confusion matrix drawing script (#572)\r\n07. Add single channel application case (#460)\r\n08. Support auto registration (#597)\r\n09. Support Box CAM of YOLOv7, YOLOv8 and PPYOLOE (#601)\r\n10. Add automated generation of MM series repo registration information and tools scripts (#559)\r\n11. Added YOLOv7 model structure diagram (#504)\r\n12. Add how to specify specific GPU training and inference files (#503)\r\n13. Add check if `metainfo` is all lowercase when training or testing (#535)\r\n14. Add links to Twitter, Discord, Medium, YouTube, etc. (#555)\r\n\r\n### Bug Fixes\r\n\r\n1. Fix isort version issue (#492, #497)\r\n2. Fix type error of assigner visualization (#509)\r\n3. Fix YOLOv8 documentation link error (#517)\r\n4. Fix RTMDet Decoder error in EasyDeploy (#519)\r\n5. Fix some document linking errors (#537)\r\n6. Fix RTMDet-Tiny weight path error (#580)\r\n\r\n### Improvements\r\n\r\n1. Update `contributing.md`\r\n2. Optimize `DetDataPreprocessor` branch to support multitasking (#511)\r\n3. Optimize `gt_instances_preprocess` so it can be used for other YOLO algorithms (#532)\r\n4. Add `yolov7-e6e` weight conversion script (#570)\r\n5. Reference YOLOv8 inference code modification PPYOLOE\r\n\r\n### Contributors\r\n\r\nA total of 23 developers contributed to this release.\r\n\r\nThank @triple-Mu, @isLinXu, @Audrey528, @TianWen580, @yechenzhi, @RangeKing, @lyviva, @Nioolek, @PeterH0323, @tianleiSHI, @aptsunny, @satuoqaq, @vansin, @xin-li-67, @VoyagerXvoyagerx,\r\n@landhill, @kitecats, @tang576225574, @HIT-cwh, @AI-Tianlong, @RangiLyu, @hhaAndroid, @liuyanyi ","2023-03-02T09:40:17",{"id":175,"version":176,"summary_zh":177,"released_at":178},105567,"v0.4.0","\u003Cimg width=\"100%\" src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F27466624\u002F213130448-1f8529fd-2247-4ac4-851c-acd0148a49b9.png\"\u002F>\r\n\r\n## v0.4.0 (18\u002F1\u002F2023)\r\n\r\n### Highlights\r\n\r\n1. Implemented [YOLOv8](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Fyolov8\u002FREADME.md) object detection model, and supports model deployment in [projects\u002Feasydeploy](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fprojects\u002Feasydeploy)\r\n2. Added Chinese and English versions of [Algorithm principles and implementation with YOLOv8](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fdocs\u002Fen\u002Falgorithm_descriptions\u002Fyolov8_description.md)\r\n\r\n### New Features\r\n\r\n1. Added YOLOv8 and PPYOLOE model structure diagrams (#459, #471)\r\n2. Adjust the minimum supported Python version from 3.6 to 3.7 (#449)\r\n3. Added a new YOLOX decoder in TensorRT-8 (#450)\r\n4. Add a tool for scheduler visualization (#479)\r\n\r\n### Bug Fixes\r\n\r\n1. Fix `optimize_anchors.py` script import error (#452)\r\n2. Fix the wrong installation steps in `get_started.md` (#474)\r\n3. Fix the neck error when using the `RTMDet` P6 model (#480)\r\n\r\n### Contributors\r\n\r\nA total of 9 developers contributed to this release.\r\n\r\nThank @VoyagerXvoyagerx, @tianleiSHI, @RangeKing, @PeterH0323, @Nioolek, @triple-Mu, @lyviva, @Zheng-LinXiao, @hhaAndroid","2023-01-18T11:55:53",{"id":180,"version":181,"summary_zh":182,"released_at":183},105568,"v0.3.0","## v0.3.0 (8\u002F1\u002F2023)\r\n\r\n### Highlights\r\n\r\n1. Implement fast version of [RTMDet](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Frtmdet\u002FREADME.md). RTMDet-s 8xA100 training takes only 14 hours. The training speed is 2.6 times faster than the previous version.\r\n2. Support [PPYOLOE](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Fppyoloe\u002FREADME.md) training\r\n3. Support `iscrowd` attribute training in [YOLOv5](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Fyolov5\u002Fcrowdhuman\u002Fyolov5_s-v61_8xb16-300e_ignore_crowdhuman.py)\r\n4. Support [YOLOv5 assigner result visualization](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fprojects\u002Fassigner_visualization\u002FREADME.md)\r\n\r\n### New Features\r\n\r\n01. Add `crowdhuman` dataset (#368)\r\n02. Easydeploy support TensorRT inference (#377)\r\n03. Add `YOLOX` structure description (#402)\r\n04. Add a feature for the video demo (#392)\r\n05. Support `YOLOv7` easy deploy (#427)\r\n06. Add resume from specific checkpoint in CLI (#393)\r\n07. Set `metainfo` fields to lower case (#362, #412)\r\n08. Add module combination doc (#349, #352, #345)\r\n09. Add docs about how to freeze the weight of backbone or neck (#418)\r\n10. Add don't used pre-training weights doc in `how_to.md` (#404)\r\n11. Add docs about how to set the random seed (#386)\r\n12. Translate `rtmdet_description.md` document to English (#353)\r\n13. Add doc of `yolov6_description.md` (#382, #372)\r\n\r\n### Bug Fixes\r\n\r\n01. Fix bugs in the output annotation file when `--class-id-txt` is set (#430)\r\n02. Fix batch inference bug in `YOLOv5` head (#413)\r\n03. Fix typehint in some heads (#415, #416, #443)\r\n04. Fix RuntimeError of `torch.cat()` expected a non-empty list of Tensors (#376)\r\n05. Fix the device inconsistency error in `YOLOv7` training (#397)\r\n06. Fix the `scale_factor` and `pad_param` value in `LetterResize` (#387)\r\n07. Fix docstring graph rendering error of readthedocs (#400)\r\n08. Fix AssertionError when `YOLOv6` from training to val (#378)\r\n09. Fix CI error due to `np.int` and legacy builder.py (#389)\r\n10. Fix MMDeploy rewriter (#366)\r\n11. Fix MMYOLO unittest scope bug (#351)\r\n12. Fix `pad_param` error (#354)\r\n13. Fix twice head inference bug (#342)\r\n14. Fix customize dataset training (#428)\r\n\r\n### Improvements\r\n\r\n01. Update `useful_tools.md` (#384)\r\n02. update the English version of `custom_dataset.md` (#381)\r\n03. Remove context argument from the rewriter function (#395)\r\n04. deprecating `np.bool` type alias (#396)\r\n05. Add new video link for custom dataset (#365)\r\n06. Export onnx for model only (#361)\r\n07. Add MMYOLO regression test yml (#359)\r\n08. Update video tutorials in `article.md` (#350)\r\n09. Add deploy demo (#343)\r\n10. Optimize the vis results of large images in debug mode (#346)\r\n11. Improve args for `browse_dataset` and support `RepeatDataset` (#340, #338)\r\n\r\n### Contributors\r\n\r\nA total of 28 developers contributed to this release.\r\n\r\nThank @RangeKing, @PeterH0323, @Nioolek, @triple-Mu, @matrixgame2018, @xin-li-67, @tang576225574, @kitecats, @Seperendity, @diplomatist, @vaew, @wzr-skn, @VoyagerXvoyagerx, @MambaWong, @tianleiSHI, @caj-github, @zhubochao, @lvhan028, @dsghaonan, @lyviva, @yuewangg, @wang-tf, @satuoqaq, @grimoire, @RunningLeon, @hanrui1sensetime, @RangiLyu, @hhaAndroid","2023-01-08T07:50:54",{"id":185,"version":186,"summary_zh":187,"released_at":188},105569,"v0.2.0","## v0.2.0（1\u002F12\u002F2022)\r\n\r\n### Highlights\r\n\r\n1. Support [YOLOv7](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Ftree\u002Fdev\u002Fconfigs\u002Fyolov7) P5 and P6 model\r\n2. Support [YOLOv6](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fconfigs\u002Fyolov6\u002FREADME.md) ML model\r\n3. Support [Grad-Based CAM and Grad-Free CAM](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fdemo\u002Fboxam_vis_demo.py)\r\n4. Support [large image inference](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fdemo\u002Flarge_image_demo.py) based on sahi\r\n5. Add [easydeploy](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fprojects\u002Feasydeploy\u002FREADME.md) project under the projects folder\r\n6. Add [custom dataset guide](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fdocs\u002Fzh_cn\u002Fuser_guides\u002Fcustom_dataset.md)\r\n\r\n### New Features\r\n\r\n1. `browse_dataset.py` script supports visualization of original image, data augmentation and intermediate results (#304)\r\n2. Add flag to output labelme label file in `image_demo.py` (#288, #314)\r\n3. Add `labelme2coco` script (#308, #313)\r\n4. Add split COCO dataset script (#311)\r\n5. Add two examples of backbone replacement in `how-to.md` and update `plugin.md` (#291)\r\n6. Add `contributing.md` and `code_style.md` (#322)\r\n7. Add docs about how to use mim to run scripts across libraries (#321)\r\n8. Support `YOLOv5` deployment at RV1126 device (#262)\r\n\r\n### Bug Fixes\r\n\r\n1. Fix MixUp padding error (#319)\r\n2. Fix scale factor order error of `LetterResize` and `YOLOv5KeepRatioResize` (#305)\r\n3. Fix training errors of `YOLOX Nano` model (#285)\r\n4. Fix `RTMDet` deploy error (#287)\r\n5. Fix int8 deploy config (#315)\r\n6. Fix `make_stage_plugins` doc in `basebackbone` (#296)\r\n7. Enable switch to deploy when create pytorch model in deployment (#324)\r\n8. Fix some errors in `RTMDet` model graph (#317)\r\n\r\n### Improvements\r\n\r\n1. Add option of json output in `test.py` (#316)\r\n2. Add area condition in `extract_subcoco.py` script (#286)\r\n3. Deployment doc translation (#289)\r\n4. Add YOLOv6 description overview doc (#252)\r\n5. Improve `config.md` (#297, #303)\r\n   6Add mosaic9 graph in docstring  (#307)\r\n6. Improve `browse_coco_json.py` script args (#309)\r\n7. Refactor some functions in `dataset_analysis.py` to be more general (#294)\r\n\r\n#### Contributors\r\n\r\nA total of 14 developers contributed to this release.\r\n\r\nThank  @fcakyon, @matrixgame2018, @MambaWong, @imAzhou, @triple-Mu, @RangeKing, @PeterH0323, @xin-li-67, @kitecats, @hanrui1sensetime, @AllentDan, @Zheng-LinXiao, @hhaAndroid, @wanghonglie","2022-12-01T03:17:52",{"id":190,"version":191,"summary_zh":192,"released_at":193},105570,"v0.1.3","## v0.1.3（10\u002F11\u002F2022)\r\n\r\n### New Features\r\n\r\n1. Support CBAM plug-in and provide plug-in documentation (#246)\r\n2. Add YOLOv5 P6 model structure diagram and related descriptions (#273)\r\n\r\n### Bug Fixes\r\n\r\n1. Fix training failure when saving best weights based on mmengine 0.3.1\r\n2. Fix `add_dump_metric` error based on mmdet 3.0.0rc3 (#253)\r\n3. Fix backbone does not support `init_cfg` issue (#272)\r\n4. Change typing import method based on mmdet 3.0.0rc3 (#261)\r\n\r\n### Improvements\r\n\r\n1. `featmap_vis_demo` support for folder and url input (#248)\r\n2. Deploy docker file refinement (#242)\r\n\r\n#### Contributors\r\n\r\nA total of 10 developers contributed to this release.\r\n\r\nThank @kitecats, @triple-Mu, @RangeKing, @PeterH0323, @Zheng-LinXiao, @tkhe, @weikai520, @zytx121, @wanghonglie, @hhaAndroid","2022-11-10T03:26:32",{"id":195,"version":196,"summary_zh":197,"released_at":198},105571,"v0.1.2","## v0.1.2（3\u002F11\u002F2022)\r\n\r\n### Highlights\r\n\r\n1. Support [YOLOv5\u002FYOLOv6\u002FYOLOX\u002FRTMDet deployments](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002Fconfigs\u002Fdeploy) for ONNXRuntime and TensorRT\r\n2. Support [YOLOv6](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002Fconfigs\u002Fyolov6) s\u002Ft\u002Fn model training\r\n3. YOLOv5 supports [P6 model training which can input 1280-scale images](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002Fconfigs\u002Fyolov5)\r\n4. YOLOv5 supports [VOC dataset training](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002Fconfigs\u002Fyolov5\u002Fvoc)\r\n5. Support [PPYOLOE](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002Fconfigs\u002Fppyoloe) and [YOLOv7](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002Fconfigs\u002Fyolov7) model inference and official weight conversion\r\n6. Add YOLOv5 replacement [backbone tutorial](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fdev\u002Fdocs\u002Fen\u002Fadvanced_guides\u002Fhow_to.md#use-backbone-network-implemented-in-other-openmmlab-repositories) in How-to documentation\r\n\r\n### New Features\r\n\r\n1. Add `optimize_anchors` script (#175)\r\n2. Add `extract_subcoco` script (#186)\r\n3. Add `yolo2coco` conversion script (#161)\r\n4. Add `dataset_analysis` script (#172)\r\n5. Remove Albu version restrictions (#187)\r\n\r\n### Bug Fixes\r\n\r\n1. Fix the problem that `cfg.resume` does not work when set (#221)\r\n2. Fix the problem of not showing bbox in feature map visualization script (#204)\r\n3. Update the metafile of RTMDet (#188)\r\n4. Fix a visualization error in `test_pipeline` (#166)\r\n5. Update badges (#140)\r\n\r\n### Improvements\r\n\r\n1. Optimize Readthedoc display page (#209)\r\n2. Add docstring for module structure diagram for base model (#196)\r\n3. Support for not including any instance logic in LoadAnnotations (#161)\r\n4. Update `image_demo` script to support folder and url paths (#128)\r\n5. Update pre-commit hook (#129)\r\n\r\n### Documentation\r\n\r\n1. Translate `yolov5_description.md`, `yolov5_tutorial.md` and `visualization.md` into English (#138, #198, #206)\r\n2. Add deployment-related Chinese documentation (#220)\r\n3. Update `config.md`, `faq.md` and `pull_request_template.md` (#190, #191, #200)\r\n4. Update the `article` page (#133)\r\n\r\n#### Contributors\r\n\r\nA total of 14 developers contributed to this release.\r\n\r\nThank @imAzhou, @triple-Mu, @RangeKing, @PeterH0323, @xin-li-67, @Nioolek, @kitecats, @Bin-ze, @JiayuXu0, @cydiachen, @zhiqwang, @Zheng-LinXiao, @hhaAndroid, @wanghonglie","2022-11-03T11:05:01",{"id":200,"version":201,"summary_zh":202,"released_at":203},105572,"v0.1.1","Based on MMDetection's RTMDet high precision and low latency object detection algorithm, we have also released RTMDet and provided a Chinese document on the principle and implementation of RTMDet.\r\n\r\n## Highlights\r\n\r\n1. Support [RTMDet](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmyolo\u002Fblob\u002Fmain\u002Fconfigs\u002Frtmdet)\r\n2. Support for backbone customization plugins and update How-to documentation (#75)\r\n\r\n## Bug Fixes\r\n\r\n1. Fix some documentation errors (#66, #72, #76, #83, #86)\r\n2. Fix checkpoints link error (#63)\r\n3. Fix the bug that the output of `LetterResize` does not meet the expectation when using `imscale` (#105)\r\n\r\n## Improvements\r\n\r\n1. Reducing the size of docker images (#67)\r\n2. Simplifying `Compose` Logic in `BaseMixImageTransform` (#71)\r\n3. Supports dump results in `test.py` (#84)\r\n\r\n## Contributors\r\n\r\nA total of 13 developers contributed to this release.\r\n\r\nThank @wanghonglie, @hhaAndroid, @yang-0201, @PeterH0323, @RangeKing, @satuoqaq, @Zheng-LinXiao, @xin-li-67, @suibe-qingtian, @MambaWong, @MichaelCai0912, @rimoire, @Nioolek","2022-09-29T11:13:35",{"id":205,"version":206,"summary_zh":207,"released_at":208},105573,"v0.1.0","We have released MMYOLO open source library, which is based on MMEngine, MMCV 2.x and MMDetection 3.x libraries. At present, the object detection has been realized, and it will be expanded to multi-task in the future.\r\n\r\n### Highlights\r\n\r\n1. Support YOLOv5\u002FYOLOX training, support YOLOv6 inference. Deployment will be supported soon.\r\n2. Refactored YOLOX from MMDetection to accelerate training and inference.\r\n3. Detailed introduction and advanced tutorials are provided, see the [English tutorial](https:\u002F\u002Fmmyolo.readthedocs.io\u002Fen\u002Flatest).","2022-09-21T11:36:25"]