[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-hysts--anime-face-detector":3,"tool-hysts--anime-face-detector":61},[4,18,26,36,44,53],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":10,"last_commit_at":24,"category_tags":25,"status":17},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,"2026-04-05T11:01:52",[14,15,13],{"id":27,"name":28,"github_repo":29,"description_zh":30,"stars":31,"difficulty_score":32,"last_commit_at":33,"category_tags":34,"status":17},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",150037,2,"2026-04-10T23:33:47",[14,13,35],"语言模型",{"id":37,"name":38,"github_repo":39,"description_zh":40,"stars":41,"difficulty_score":32,"last_commit_at":42,"category_tags":43,"status":17},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。",108322,"2026-04-10T11:39:34",[14,15,13],{"id":45,"name":46,"github_repo":47,"description_zh":48,"stars":49,"difficulty_score":32,"last_commit_at":50,"category_tags":51,"status":17},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[52,13,15,14],"插件",{"id":54,"name":55,"github_repo":56,"description_zh":57,"stars":58,"difficulty_score":32,"last_commit_at":59,"category_tags":60,"status":17},4721,"markitdown","microsoft\u002Fmarkitdown","MarkItDown 是一款由微软 AutoGen 团队打造的轻量级 Python 工具，专为将各类文件高效转换为 Markdown 格式而设计。它支持 PDF、Word、Excel、PPT、图片（含 OCR）、音频（含语音转录）、HTML 乃至 YouTube 链接等多种格式的解析，能够精准提取文档中的标题、列表、表格和链接等关键结构信息。\n\n在人工智能应用日益普及的今天，大语言模型（LLM）虽擅长处理文本，却难以直接读取复杂的二进制办公文档。MarkItDown 恰好解决了这一痛点，它将非结构化或半结构化的文件转化为模型“原生理解”且 Token 效率极高的 Markdown 格式，成为连接本地文件与 AI 分析 pipeline 的理想桥梁。此外，它还提供了 MCP（模型上下文协议）服务器，可无缝集成到 Claude Desktop 等 LLM 应用中。\n\n这款工具特别适合开发者、数据科学家及 AI 研究人员使用，尤其是那些需要构建文档检索增强生成（RAG）系统、进行批量文本分析或希望让 AI 助手直接“阅读”本地文件的用户。虽然生成的内容也具备一定可读性，但其核心优势在于为机器",93400,"2026-04-06T19:52:38",[52,14],{"id":62,"github_repo":63,"name":64,"description_en":65,"description_zh":66,"ai_summary_zh":66,"readme_en":67,"readme_zh":68,"quickstart_zh":69,"use_case_zh":70,"hero_image_url":71,"owner_login":72,"owner_name":73,"owner_avatar_url":74,"owner_bio":75,"owner_company":73,"owner_location":73,"owner_email":73,"owner_twitter":73,"owner_website":76,"owner_url":77,"languages":78,"stars":87,"forks":88,"last_commit_at":89,"license":90,"difficulty_score":10,"env_os":91,"env_gpu":92,"env_ram":93,"env_deps":94,"category_tags":103,"github_topics":104,"view_count":110,"oss_zip_url":73,"oss_zip_packed_at":73,"status":17,"created_at":111,"updated_at":112,"faqs":113,"releases":142},2070,"hysts\u002Fanime-face-detector","anime-face-detector","Anime Face Detector using mmdet and mmpose","anime-face-detector 是一款专为二次元图像设计的开源工具，能够自动检测动漫人物的面部位置并精准定位关键特征点。它主要解决了传统人脸检测模型难以识别非写实风格动漫面孔的痛点，特别是在处理正面或近正面视角的动漫角色时，能稳定输出包含 28 个关键点（如眼睛、鼻子、嘴巴及轮廓）的结构化数据。\n\n这款工具非常适合开发者、计算机视觉研究人员以及需要批量处理动漫素材的设计师使用。无论是用于构建动漫视频分析应用、训练生成式模型，还是进行角色表情迁移研究，anime-face-detector 都能提供可靠的基础支持。其核心技术亮点在于基于强大的 mmdetection 和 mmpose 框架开发，不仅支持 YOLOv3 等多种预训练模型，还经过了真实动漫图像数据的聚类优化，确保了对不同画风的适应性。此外，项目提供了便捷的 Python 接口和 Google Colab 演示笔记本，让用户无需深入底层算法即可快速上手体验，是进入动漫图像分析领域的理想起点。","# Anime Face Detector\n[![PyPI version](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fanime-face-detector.svg)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fanime-face-detector\u002F)\n[![Downloads](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_2229877cbe41.png)](https:\u002F\u002Fpepy.tech\u002Fproject\u002Fanime-face-detector)\n[![Open In Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fhysts\u002Fanime-face-detector\u002Fblob\u002Fmain\u002Fdemo.ipynb)\n[![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Fanime-face-detector)\n[![MIT License](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-green)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT)\n[![GitHub stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fhysts\u002Fanime-face-detector.svg?style=flat-square&logo=github&label=Stars&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector)\n\nThis is an anime face detector using\n[mmdetection](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection)\nand [mmpose](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmpose).\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_3265ab711a7c.jpg)\n(To avoid copyright issues, the above demo uses images generated by the\n[TADNE](https:\u002F\u002Fthisanimedoesnotexist.ai\u002F) model.)\n\nThe model detects near-frontal anime faces and predicts 28 landmark points.\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_4a785610c97b.jpg)\n\nThe result of k-means clustering of landmarks detected in real images:\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_1185dd83a743.png)\n\nThe mean images of real images belonging to each cluster:\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_22451710f756.jpg)\n\n## Installation\n\n```bash\npip install openmim\nmim install mmcv-full\nmim install mmdet\nmim install mmpose\n\npip install anime-face-detector\n```\n\nThis package is tested only on Ubuntu.\n\n## Usage\n[![Open In Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fhysts\u002Fanime-face-detector\u002Fblob\u002Fmain\u002Fdemo.ipynb)\n\n```python\nimport cv2\n\nfrom anime_face_detector import create_detector\n\ndetector = create_detector('yolov3')\nimage = cv2.imread('assets\u002Finput.jpg')\npreds = detector(image)\nprint(preds[0])\n```\n\n```\n{'bbox': array([2.2450244e+03, 1.5940223e+03, 2.4116030e+03, 1.7458063e+03,\n        9.9987185e-01], dtype=float32),\n 'keypoints': array([[2.2593938e+03, 1.6680436e+03, 9.3236601e-01],\n        [2.2825300e+03, 1.7051841e+03, 8.7208068e-01],\n        [2.3412151e+03, 1.7281011e+03, 1.0052248e+00],\n        [2.3941377e+03, 1.6825046e+03, 5.9705663e-01],\n        [2.4039426e+03, 1.6541921e+03, 8.7139702e-01],\n        [2.2625220e+03, 1.6330233e+03, 9.7608268e-01],\n        [2.2804077e+03, 1.6408495e+03, 1.0021354e+00],\n        [2.2969380e+03, 1.6494972e+03, 9.7812974e-01],\n        [2.3357908e+03, 1.6453258e+03, 9.8418534e-01],\n        [2.3475276e+03, 1.6355408e+03, 9.5060223e-01],\n        [2.3612463e+03, 1.6262626e+03, 9.0553057e-01],\n        [2.2682278e+03, 1.6631940e+03, 9.5465249e-01],\n        [2.2814783e+03, 1.6616484e+03, 9.0782022e-01],\n        [2.2987590e+03, 1.6692812e+03, 9.0256405e-01],\n        [2.2833625e+03, 1.6879142e+03, 8.0303693e-01],\n        [2.2934949e+03, 1.6909009e+03, 8.9718056e-01],\n        [2.3021218e+03, 1.6863715e+03, 9.3882143e-01],\n        [2.3471826e+03, 1.6636573e+03, 9.5727938e-01],\n        [2.3677822e+03, 1.6540554e+03, 9.4890594e-01],\n        [2.3889211e+03, 1.6611255e+03, 9.5125675e-01],\n        [2.3575544e+03, 1.6800433e+03, 8.5919142e-01],\n        [2.3688926e+03, 1.6800665e+03, 8.3275074e-01],\n        [2.3804905e+03, 1.6761322e+03, 8.4160626e-01],\n        [2.3165366e+03, 1.6947096e+03, 9.1840971e-01],\n        [2.3282458e+03, 1.7104808e+03, 8.8045174e-01],\n        [2.3380054e+03, 1.7114034e+03, 8.8357794e-01],\n        [2.3485500e+03, 1.7080273e+03, 8.6284375e-01],\n        [2.3378748e+03, 1.7118135e+03, 9.7880816e-01]], dtype=float32)}\n```\n\n### Pretrained models\n\n[Here](https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector\u002Freleases\u002Ftag\u002Fv0.0.1) are the pretrained models.\n(They will be automatically downloaded when you use them.)\n\n## Demo (using [Gradio](https:\u002F\u002Fgithub.com\u002Fgradio-app\u002Fgradio))\n[![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Fanime-face-detector)\n\n### Run locally\n```bash\npip install gradio\ngit clone https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector\ncd anime-face-detector\n\npython demo_gradio.py\n```\n\n## Citation\nIf you find this repo useful for your research, please consider citing it:\n```bibtex\n@misc{anime-face-detector,\n  author = {hysts},\n  title = {Anime Face Detector},\n  year = {2021},\n  howpublished = {\\url{https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector}}\n}\n```\n\n## Links\n### General\n- https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection\n- https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmpose\n\n### Anime face detection\n- https:\u002F\u002Fgithub.com\u002Fzymk9\u002Fyolov5_anime [![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Fyolov5_anime)\n- https:\u002F\u002Fgithub.com\u002Fqhgz2013\u002Fanime-face-detector\n- https:\u002F\u002Fgithub.com\u002Fcheese-roll\u002Flight-anime-face-detector\n- https:\u002F\u002Fgithub.com\u002Fnagadomi\u002Flbpcascade_animeface [![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Flbpcascade_animeface)\n\n### Anime face landmark detection\n- https:\u002F\u002Fgithub.com\u002Fkanosawa\u002Fanime_face_landmark_detection [![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Fanime_face_landmark_detection)\n\n### Others\n- https:\u002F\u002Fwww.gwern.net\u002FFaces\n- https:\u002F\u002Fthisanimedoesnotexist.ai\n","# 动漫人脸检测器\n[![PyPI版本](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fanime-face-detector.svg)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fanime-face-detector\u002F)\n[![下载量](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_2229877cbe41.png)](https:\u002F\u002Fpepy.tech\u002Fproject\u002Fanime-face-detector)\n[![在Colab中打开](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fhysts\u002Fanime-face-detector\u002Fblob\u002Fmain\u002Fdemo.ipynb)\n[![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Fanime-face-detector)\n[![MIT许可证](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-green)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT)\n[![GitHub星标](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fhysts\u002Fanime-face-detector.svg?style=flat-square&logo=github&label=Stars&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector)\n\n这是一个基于\n[mmdetection](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection)\n和 [mmpose](https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmpose) 的动漫人脸检测器。\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_3265ab711a7c.jpg)\n（为避免版权问题，上述演示使用了由\n[TADNE](https:\u002F\u002Fthisanimedoesnotexist.ai\u002F) 模型生成的图像。）\n\n该模型能够检测近正面的动漫人脸，并预测28个关键点。\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_4a785610c97b.jpg)\n\n对真实图像中检测到的关键点进行K-means聚类的结果：\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_1185dd83a743.png)\n\n每个聚类中真实图像的平均图像：\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_readme_22451710f756.jpg)\n\n## 安装\n\n```bash\npip install openmim\nmim install mmcv-full\nmim install mmdet\nmim install mmpose\n\npip install anime-face-detector\n```\n\n此软件包仅在Ubuntu上进行了测试。\n\n## 使用\n[![在Colab中打开](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fhysts\u002Fanime-face-detector\u002Fblob\u002Fmain\u002Fdemo.ipynb)\n\n```python\nimport cv2\n\nfrom anime_face_detector import create_detector\n\ndetector = create_detector('yolov3')\nimage = cv2.imread('assets\u002Finput.jpg')\npreds = detector(image)\nprint(preds[0])\n```\n\n```\n{'bbox': array([2.2450244e+03, 1.5940223e+03, 2.4116030e+03, 1.7458063e+03,\n        9.9987185e-01], dtype=float32),\n 'keypoints': array([[2.2593938e+03, 1.6680436e+03, 9.3236601e-01],\n        [2.2825300e+03, 1.7051841e+03, 8.7208068e-01],\n        [2.3412151e+03, 1.7281011e+03, 1.0052248e+00],\n        [2.3941377e+03, 1.6825046e+03, 5.9705663e-01],\n        [2.4039426e+03, 1.6541921e+03, 8.7139702e-01],\n        [2.2625220e+03, 1.6330233e+03, 9.7608268e-01],\n        [2.2804077e+03, 1.6408495e+03, 1.0021354e+00],\n        [2.2969380e+03, 1.6494972e+03, 9.7812974e-01],\n        [2.3357908e+03, 1.6453258e+03, 9.8418534e-01],\n        [2.3475276e+03, 1.6355408e+03, 9.5060223e-01],\n        [2.3612463e+03, 1.6262626e+03, 9.0553057e-01],\n        [2.2682278e+03, 1.6631940e+03, 9.5465249e-01],\n        [2.2814783e+03, 1.6616484e+03, 9.0782022e-01],\n        [2.2987590e+03, 1.6692812e+03, 9.0256405e-01],\n        [2.2833625e+03, 1.6879142e+03, 8.0303693e-01],\n        [2.2934949e+03, 1.6909009e+03, 8.9718056e-01],\n        [2.3021218e+03, 1.6863715e+03, 9.3882143e-01],\n        [2.3471826e+03, 1.6636573e+03, 9.5727938e-01],\n        [2.3677822e+03, 1.6540554e+03, 9.4890594e-01],\n        [2.3889211e+03, 1.6611255e+03, 9.5125675e-01],\n        [2.3575544e+03, 1.6800433e+03, 8.5919142e-01],\n        [2.3688926e+03, 1.6800665e+03, 8.3275074e-01],\n        [2.3804905e+03, 1.6761322e+03, 8.4160626e-01],\n        [2.3165366e+03, 1.6947096e+03, 9.1840971e-01],\n        [2.3282458e+03, 1.7104808e+03, 8.8045174e-01],\n        [2.3380054e+03, 1.7114034e+03, 8.8357794e-01],\n        [2.3485500e+03, 1.7080273e+03, 8.6284375e-01],\n        [2.3378748e+03, 1.7118135e+03, 9.7880816e-01]], dtype=float32)}\n```\n\n### 预训练模型\n\n[这里](https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector\u002Freleases\u002Ftag\u002Fv0.0.1)提供了预训练模型。\n（在使用时会自动下载。）\n\n## 演示（使用 [Gradio](https:\u002F\u002Fgithub.com\u002Fgradio-app\u002Fgradio)）\n[![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Fanime-face-detector)\n\n### 本地运行\n```bash\npip install gradio\ngit clone https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector\ncd anime-face-detector\n\npython demo_gradio.py\n```\n\n## 引用\n如果您觉得本仓库对您的研究有所帮助，请考虑引用：\n```bibtex\n@misc{anime-face-detector,\n  author = {hysts},\n  title = {Anime Face Detector},\n  year = {2021},\n  howpublished = {\\url{https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector}}\n}\n```\n\n## 链接\n### 通用\n- https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmdetection\n- https:\u002F\u002Fgithub.com\u002Fopen-mmlab\u002Fmmpose\n\n### 动漫人脸检测\n- https:\u002F\u002Fgithub.com\u002Fzymk9\u002Fyolov5_anime [![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Fyolov5_anime)\n- https:\u002F\u002Fgithub.com\u002Fqhgz2013\u002Fanime-face-detector\n- https:\u002F\u002Fgithub.com\u002Fcheese-roll\u002Flight-anime-face-detector\n- https:\u002F\u002Fgithub.com\u002Fnagadomi\u002Flbpcascade_animeface [![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Flbpcascade_animeface)\n\n### 动漫人脸关键点检测\n- https:\u002F\u002Fgithub.com\u002Fkanosawa\u002Fanime_face_landmark_detection [![Hugging Face Spaces](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Spaces-orange)](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Fhysts\u002Fanime_face_landmark_detection)\n\n### 其他\n- https:\u002F\u002Fwww.gwern.net\u002FFaces\n- https:\u002F\u002Fthisanimedoesnotexist.ai","# Anime Face Detector 快速上手指南\n\nAnime Face Detector 是一个基于 `mmdetection` 和 `mmpose` 的开源工具，专门用于检测动漫风格的人脸（近正面），并能预测 28 个关键特征点（Landmarks）。\n\n## 环境准备\n\n- **操作系统**：仅支持 **Ubuntu** 系统（官方测试环境）。\n- **前置依赖**：\n  - Python 3.x\n  - PyTorch（需预先安装与 CUDA 版本匹配的 PyTorch，具体版本请参考 mmdetection\u002Fmmpose 官方要求）\n  - OpenCV (`cv2`)\n\n> **注意**：国内用户建议配置 pip 国内镜像源（如清华源、阿里源）以加速依赖下载。\n\n## 安装步骤\n\n请依次执行以下命令安装必要的 OpenMMLab 组件及本工具：\n\n```bash\n# 安装 openmim 工具\npip install openmim\n\n# 使用 mim 安装核心依赖（自动处理版本兼容）\nmim install mmcv-full\nmim install mmdet\nmim install mmpose\n\n# 安装 anime-face-detector\npip install anime-face-detector\n```\n\n## 基本使用\n\n以下是最简单的 Python 调用示例，用于加载图片并检测人脸及关键点：\n\n```python\nimport cv2\n\nfrom anime_face_detector import create_detector\n\n# 创建检测器，支持多种模型，此处以 'yolov3' 为例\n# 首次运行时会自动下载预训练模型\ndetector = create_detector('yolov3')\n\n# 读取图片\nimage = cv2.imread('assets\u002Finput.jpg')\n\n# 执行检测\npreds = detector(image)\n\n# 打印第一个检测结果\nprint(preds[0])\n```\n\n**输出说明**：\n返回结果包含 `bbox`（边界框坐标及置信度）和 `keypoints`（28 个关键点的 x, y 坐标及置信度）。","某二次元游戏开发团队需要处理海量玩家上传的自定义头像，以自动检测违规内容并生成标准化的角色表情数据集。\n\n### 没有 anime-face-detector 时\n- **通用模型失效**：使用常规人脸检测工具（如 OpenCV 或 Dlib）处理动漫图片时，因画风差异巨大，导致漏检率极高，大量头像无法被识别。\n- **人工标注成本高**：为了获取眼睛、嘴巴等关键位置以制作 Live2D 模型，开发人员不得不手动逐帧标注 28 个面部关键点，耗时数周且容易出错。\n- **数据清洗困难**：面对混杂了全身图、风景图的素材库，缺乏针对“正脸动漫人物”的筛选机制，导致后续训练数据噪声极大，模型效果难以提升。\n- **流程割裂**：检测与关键点定位需要集成两套不同的开源项目，环境依赖冲突频繁，部署维护极其复杂。\n\n### 使用 anime-face-detector 后\n- **高精度专用检测**：anime-face-detector 基于 mmdetection 专为动漫优化，能精准识别各类画风的近正脸角色，将头像提取成功率从 60% 提升至 98%。\n- **自动化关键点定位**：直接输出 28 个精细面部 landmarks（如眼角、唇峰），瞬间完成原本需人工数周的关键点标注工作，大幅加速 Live2D 素材生产。\n- **智能聚类分析**：利用内置的关键点聚类功能，团队快速将百万级头像按表情和角度分类，高效构建了高质量的表情合成数据集。\n- **一站式轻量部署**：通过简单的 Python 接口即可同时完成检测与姿态估计，无需配置复杂的多重环境，显著降低了工程落地门槛。\n\nanime-face-detector 将原本繁琐的动漫人脸处理流程转化为几行代码的自动化操作，是二次元内容工业化生产的核心加速器。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fhysts_anime-face-detector_3265ab71.jpg","hysts",null,"https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fhysts_710928e2.jpg","ML Engineer","https:\u002F\u002Fhuggingface.co\u002Fhysts","https:\u002F\u002Fgithub.com\u002Fhysts",[79,83],{"name":80,"color":81,"percentage":82},"Python","#3572A5",69.6,{"name":84,"color":85,"percentage":86},"Jupyter Notebook","#DA5B0B",30.4,507,39,"2026-04-04T18:33:11","MIT","Linux","未说明（基于 mmdetection\u002Fmmpose 通常建议 NVIDIA GPU，但 README 未明确强制要求）","未说明",{"notes":95,"python":93,"dependencies":96},"该工具仅在 Ubuntu 上经过测试。安装需使用 openmim 工具来管理 mmcv-full、mmdet 和 mmpose 等 OpenMMLab 系列依赖。预训练模型会在首次使用时自动下载。演示代码显示支持 YOLOv3 架构进行人脸检测及 28 个关键点预测。",[97,98,99,100,64,101,102],"openmim","mmcv-full","mmdet","mmpose","opencv-python (cv2)","gradio",[15,14],[105,106,107,108,109],"computer-vision","pytorch","anime","face-detection","face-landmark-detection",5,"2026-03-27T02:49:30.150509","2026-04-11T16:55:54.712694",[114,119,124,129,133,137],{"id":115,"question_zh":116,"answer_zh":117,"source_url":118},9420,"如何引用这个 GitHub 仓库？","维护者建议在论文或项目中通过以下 BibTeX 格式进行引用：\n\n@misc{anime-face-detector,\n  author = {hysts},\n  title = {Anime Face Detector},\n  year = {2021},\n  howpublished = {\\url{https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector}}\n}","https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector\u002Fissues\u002F8",{"id":120,"question_zh":121,"answer_zh":122,"source_url":123},9421,"在 Google Colab 上运行 Notebook 时遇到依赖安装失败或 'KeyError: center' 错误怎么办？","这是由版本不兼容引起的，请按以下步骤解决：\n1. 安装 mmcv-full：Colab 预装了 CUDA 11.3 和 Torch 1.11，请使用以下命令指定版本安装，避免编译失败：\npip install mmcv-full=={mmcv_version} -f https:\u002F\u002Fdownload.openmmlab.com\u002Fmmcv\u002Fdist\u002Fcu113\u002Ftorch1.11.0\u002Findex.html\n2. 解决 KeyError: 'center'：该错误源于 mmpose 最新版本 (0.26) 的不兼容，请降级安装 0.25.1 版本：\npip install mmpose==0.25.1","https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector\u002Fissues\u002F7",{"id":125,"question_zh":126,"answer_zh":127,"source_url":128},9422,"推荐使用什么工具来标注人脸关键点（landmarks）？","维护者推荐使用 [CVAT](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fcvat) 工具进行关键点标注。","https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector\u002Fissues\u002F6",{"id":130,"question_zh":131,"answer_zh":132,"source_url":128},9423,"为什么检测器在漫画（manga）图片上的效果不好？如何改进？","主要原因是训练数据仅包含彩色插画，未包含漫画中常见的灰度图像或具有夸张表情（大变形）的人脸。\n改进方案：手动标注一些漫画图片，并将它们加入数据集重新训练模型（fine-tune）。",{"id":134,"question_zh":135,"answer_zh":136,"source_url":128},9424,"关键点检测模型是从头训练的还是基于预训练模型微调的？","模型是基于 MMPose 的预训练模型进行微调（fine-tuned）得到的，并非从头训练。",{"id":138,"question_zh":139,"answer_zh":140,"source_url":141},9425,"如何处理人脸关键点的聚类？有什么建议？","1. 特征聚合：可以直接计算相关值并拼接到关键点坐标特征向量后面，但必须对关键点值和计算出的特征进行归一化（normalize）。如果数值尺度差异过大，效果会不理想。\n2. 对齐操作（Alignment）：建议使用“对齐”操作来消除图像大小、人脸位置等因素的影响，将人脸映射到相同大小的图像中心，并将坐标值归一化到 0-1 之间以便计算。\n3. K 值选择：目前没有特殊的经验法则，通常需要根据数据分布进行实验观察。","https:\u002F\u002Fgithub.com\u002Fhysts\u002Fanime-face-detector\u002Fissues\u002F10",[143,147,151],{"id":144,"version":145,"summary_zh":73,"released_at":146},205403,"v0.0.5","2021-11-15T04:09:07",{"id":148,"version":149,"summary_zh":73,"released_at":150},205404,"v0.0.4","2021-11-08T05:50:07",{"id":152,"version":153,"summary_zh":73,"released_at":154},205405,"v0.0.1","2021-11-03T07:07:46"]