[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-ipazc--mtcnn":3,"tool-ipazc--mtcnn":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":80,"owner_email":81,"owner_twitter":81,"owner_website":82,"owner_url":83,"languages":84,"stars":93,"forks":94,"last_commit_at":95,"license":96,"difficulty_score":97,"env_os":98,"env_gpu":98,"env_ram":98,"env_deps":99,"category_tags":104,"github_topics":105,"view_count":10,"oss_zip_url":81,"oss_zip_packed_at":81,"status":16,"created_at":113,"updated_at":114,"faqs":115,"releases":146},11632,"ipazc\u002Fmtcnn","mtcnn","MTCNN face detection implementation for TensorFlow, as a PIP package.","MTCNN 是一个基于 TensorFlow 的人脸检测和对齐工具，适合需要在图片中定位人脸并识别关键点的开发者和研究人员使用。它通过多任务级联卷积神经网络（Multitask Cascaded Convolutional Networks）实现高效的人脸检测和特征点定位，能够输出人脸边界框以及眼睛、鼻子、嘴巴等五个关键点的位置信息。\n\n这一工具解决了传统人脸检测方法效率低、精度不足的问题，尤其在复杂背景下仍能保持较高的检测准确率。MTCNN 采用三级网络结构（PNet、RNet 和 ONet），逐步筛选和优化候选区域，确保检测结果既快速又精确。此外，它支持批量处理，简化了大规模数据集的操作流程，并针对性能进行了优化，避免了不必要的计算开销。\n\n对于希望快速集成人脸检测功能的开发者来说，MTCNN 提供了简单易用的 Python 接口，只需几行代码即可完成安装和调用。同时，由于其开源特性，研究人员也可以深入研究其实现细节或进行二次开发。无论是构建人脸识别系统、美颜应用还是其他涉及人脸分析的项目，MTCNN 都是一个可靠的选择。","# MTCNN - Multitask Cascaded Convolutional Networks for Face Detection and Alignment\n\n[![PyPI version](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fmtcnn.svg)](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fmtcnn)\n[![Documentation Status](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fipazc_mtcnn_readme_13d664e1afd7.png)](https:\u002F\u002Fmtcnn.readthedocs.io\u002Fen\u002Flatest\u002F?badge=latest)\n![Test Status](https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Factions\u002Fworkflows\u002Ftests.yml\u002Fbadge.svg)\n![Pylint Check](https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Factions\u002Fworkflows\u002Fpylint.yml\u002Fbadge.svg)\n![PyPI Downloads](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fmtcnn)\n[![DOI](https:\u002F\u002Fzenodo.org\u002Fbadge\u002FDOI\u002F10.5281\u002Fzenodo.13901378.svg)](https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.13901378)\n\n\n\n## Overview\n\n![Example](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fipazc_mtcnn_readme_87a279b916fd.jpg)\n\nMTCNN is a robust face detection and alignment library implemented for Python >= 3.10 and TensorFlow >= 2.12, designed to detect faces and their landmarks using a multitask cascaded convolutional network. This library improves on the original implementation by offering a complete refactor, simplifying usage, improving performance, and providing support for batch processing.\n\nThis library is ideal for applications requiring face detection and alignment, with support for both bounding box and landmark prediction.\n\n## Installation\n\nMTCNN can be installed via pip:\n\n```bash\npip install mtcnn\n```\n\nMTCNN requires Tensorflow >= 2.12. This external dependency can be installed manually or automatically along with MTCNN via:\n\n```bash\npip install mtcnn[tensorflow]\n```\n\n## Usage Example\n\n```python\nfrom mtcnn import MTCNN\nfrom mtcnn.utils.images import load_image\n\n# Create a detector instance\ndetector = MTCNN(device=\"CPU:0\")\n\n# Load an image\nimage = load_image(\"ivan.jpg\")\n\n# Detect faces in the image\nresult = detector.detect_faces(image)\n\n# Display the result\nprint(result)\n```\n\nOutput example:\n\n```json\n[\n    {\n        \"box\": [277, 90, 48, 63],\n        \"keypoints\": {\n            \"nose\": [303, 131],\n            \"mouth_right\": [313, 141],\n            \"right_eye\": [314, 114],\n            \"left_eye\": [291, 117],\n            \"mouth_left\": [296, 143]\n        },\n        \"confidence\": 0.9985\n    }\n]\n```\n\n## Models Overview\n\nMTCNN uses a cascade of three networks to detect faces and facial landmarks:\n\n- **PNet (Proposal Network)**: Scans the image and proposes candidate face regions. \n- **RNet (Refine Network)**: Refines the face proposals from PNet.\n- **ONet (Output Network)**: Detects facial landmarks (eyes, nose, mouth) and provides a final refinement of the bounding boxes.\n\nAll networks are implemented using TensorFlow’s functional API and optimized to avoid unnecessary operations, such as transpositions, ensuring faster and more efficient execution.\n\n# Documentation\n\nThe full documentation for this project is available at [Read the Docs](http:\u002F\u002Fmtcnn.readthedocs.io\u002F).\n\n\n## Citation\n\nIf you use this library implementation for your research or projects, please consider using this cite:\n\n```\n@software{ivan_de_paz_centeno_2024_13901378,\n  author       = {Iván de Paz Centeno},\n  title        = {ipazc\u002Fmtcnn: v1.0.0},\n  month        = oct,\n  year         = 2024,\n  publisher    = {Zenodo},\n  version      = {v1.0.0},\n  doi          = {10.5281\u002Fzenodo.13901378},\n  url          = {https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.13901378}\n}\n```\n\nAnd the original research work from Kaipeng Zhang:\n\n```\n@article{7553523,\n    author={K. Zhang and Z. Zhang and Z. Li and Y. Qiao}, \n    journal={IEEE Signal Processing Letters}, \n    title={Joint Face Detection and Alignment Using Multitask Cascaded Convolutional Networks}, \n    year={2016}, \n    volume={23}, \n    number={10}, \n    pages={1499-1503}, \n    keywords={Benchmark testing;Computer architecture;Convolution;Detectors;Face;Face detection;Training;Cascaded convolutional neural network (CNN);face alignment;face detection}, \n    doi={10.1109\u002FLSP.2016.2603342}, \n    ISSN={1070-9908}, \n    month={Oct}\n}\n```\n\nYou may also reference the original GitHub repository that this project was based on (including the networks weights):  \n[Original MTCNN Implementation by Kaipeng Zhang](https:\u002F\u002Fgithub.com\u002Fkpzhang93\u002FMTCNN_face_detection_alignment\u002Ftree\u002Fmaster\u002Fcode)\n\nAnd the FaceNet's implementation that served as inspiration:\n[Facenet's MTCNN implementation](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Ftree\u002Fmaster\u002Fsrc\u002Falign)\n\n\n## About the Author\n\nThis project is developed and maintained by [Iván de Paz Centeno](https:\u002F\u002Fipazc.com), with the goal of standardizing face detection and providing an easy-to-use framework to help the research community push the boundaries of AI knowledge.\n\nIf you find this project useful, please consider supporting it through GitHub Sponsors. Your support will help cover costs related to improving the codebase, adding new features, and providing better documentation.\n\n[![Sponsor](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FSponsor-GitHub%20Sponsors-brightgreen)](https:\u002F\u002Fgithub.com\u002Fsponsors\u002Fipazc)\n\n\n## Acknowledgments\n\nThis project has evolved over time with contributions from multiple developers. While the current codebase has been completely rewritten, we acknowledge and appreciate the valuable input and collaboration from past contributors.\n\nA special thanks to everyone who has submitted pull requests, reported issues, or provided feedback to make this project better. \n\nFor a full list of contributors, please visit the [GitHub contributors page](https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fgraphs\u002Fcontributors).\n\n\n## License\n\nThis project is licensed under the [MIT License](LICENSE).\n","# MTCNN - 用于人脸检测和对齐的多任务级联卷积网络（Multitask Cascaded Convolutional Networks）\n\n[![PyPI version](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fmtcnn.svg)](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fmtcnn)\n[![Documentation Status](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fipazc_mtcnn_readme_13d664e1afd7.png)](https:\u002F\u002Fmtcnn.readthedocs.io\u002Fen\u002Flatest\u002F?badge=latest)\n![Test Status](https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Factions\u002Fworkflows\u002Ftests.yml\u002Fbadge.svg)\n![Pylint Check](https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Factions\u002Fworkflows\u002Fpylint.yml\u002Fbadge.svg)\n![PyPI Downloads](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fmtcnn)\n[![DOI](https:\u002F\u002Fzenodo.org\u002Fbadge\u002FDOI\u002F10.5281\u002Fzenodo.13901378.svg)](https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.13901378)\n\n\n\n## 概述\n\n![Example](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fipazc_mtcnn_readme_87a279b916fd.jpg)\n\nMTCNN 是一个健壮的人脸检测和对齐库，专为 Python（一种高级编程语言）>= 3.10 和 TensorFlow（一个开源的机器学习框架）>= 2.12 实现，旨在使用多任务级联卷积网络检测人脸及其关键点（landmark）。该库通过提供完整重构、简化使用流程、提升性能以及支持批量处理，改进了原始实现。\n\n该库适用于需要人脸检测和对齐的应用程序，支持边界框（bounding box）和关键点（landmark）预测。\n\n## 安装\n\nMTCNN 可以通过 pip 安装：\n\n```bash\npip install mtcnn\n```\n\nMTCNN 需要 TensorFlow >= 2.12。此外部依赖可手动安装，或通过以下命令与 MTCNN 一起自动安装：\n\n```bash\npip install mtcnn[tensorflow]\n```\n\n## 使用示例\n\n```python\nfrom mtcnn import MTCNN\nfrom mtcnn.utils.images import load_image\n\n# Create a detector instance\ndetector = MTCNN(device=\"CPU:0\")\n\n# Load an image\nimage = load_image(\"ivan.jpg\")\n\n# Detect faces in the image\nresult = detector.detect_faces(image)\n\n# Display the result\nprint(result)\n```\n\n输出示例：\n\n```json\n[\n    {\n        \"box\": [277, 90, 48, 63],\n        \"keypoints\": {\n            \"nose\": [303, 131],\n            \"mouth_right\": [313, 141],\n            \"right_eye\": [314, 114],\n            \"left_eye\": [291, 117],\n            \"mouth_left\": [296, 143]\n        },\n        \"confidence\": 0.9985\n    }\n]\n```\n\n## 模型概述\n\nMTCNN 使用三个网络的级联来检测人脸和面部关键点：\n\n- **PNet (Proposal Network，建议网络)**: 扫描图像并提出候选人脸区域。 \n- **RNet (Refine Network，精炼网络)**: 精炼来自 PNet 的人脸建议。\n- **ONet (Output Network，输出网络)**: 检测面部关键点（眼睛、鼻子、嘴巴）并提供边界框的最终精炼。\n\n所有网络均使用 TensorFlow 的函数式 API 实现，并经过优化以避免不必要的操作（如转置），确保更快、更高效的执行。\n\n# 文档\n\n该项目的完整文档可在 [Read the Docs](http:\u002F\u002Fmtcnn.readthedocs.io\u002F) 获取。\n\n\n## 引用\n\n如果您在研究或项目中使用此库实现，请考虑引用以下内容：\n\n```\n@software{ivan_de_paz_centeno_2024_13901378,\n  author       = {Iván de Paz Centeno},\n  title        = {ipazc\u002Fmtcnn: v1.0.0},\n  month        = oct,\n  year         = 2024,\n  publisher    = {Zenodo},\n  version      = {v1.0.0},\n  doi          = {10.5281\u002Fzenodo.13901378},\n  url          = {https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.13901378}\n}\n```\n\n以及 Kaipeng Zhang 的原始研究工作：\n\n```\n@article{7553523,\n    author={K. Zhang and Z. Zhang and Z. Li and Y. Qiao}, \n    journal={IEEE Signal Processing Letters}, \n    title={Joint Face Detection and Alignment Using Multitask Cascaded Convolutional Networks}, \n    year={2016}, \n    volume={23}, \n    number={10}, \n    pages={1499-1503}, \n    keywords={Benchmark testing;Computer architecture;Convolution;Detectors;Face;Face detection;Training;Cascaded convolutional neural network (CNN);face alignment;face detection}, \n    doi={10.1109\u002FLSP.2016.2603342}, \n    ISSN={1070-9908}, \n    month={Oct}\n}\n```\n\n您也可以引用此项目所基于的原始 GitHub 仓库（包括网络权重）：  \n[Original MTCNN Implementation by Kaipeng Zhang](https:\u002F\u002Fgithub.com\u002Fkpzhang93\u002FMTCNN_face_detection_alignment\u002Ftree\u002Fmaster\u002Fcode)\n\n以及作为灵感来源的 FaceNet 实现：  \n[Facenet's MTCNN implementation](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Ftree\u002Fmaster\u002Fsrc\u002Falign)\n\n\n## 关于作者\n\n本项目由 [Iván de Paz Centeno](https:\u002F\u002Fipazc.com) 开发和维护，旨在标准化人脸检测并提供一个易于使用的框架，以帮助研究社区推动人工智能知识的边界。\n\n如果您发现此项目有用，请考虑通过 GitHub Sponsors 支持它。您的支持将有助于覆盖改进代码库、添加新功能和提供更好文档的相关成本。\n\n[![Sponsor](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FSponsor-GitHub%20Sponsors-brightgreen)](https:\u002F\u002Fgithub.com\u002Fsponsors\u002Fipazc)\n\n\n## 致谢\n\n本项目随着时间的推移，得益于多位开发者的贡献而不断发展。尽管当前代码库已完全重写，但我们承认并感谢过去贡献者的宝贵输入和合作。\n\n特别感谢所有提交拉取请求、报告问题或提供反馈以使本项目更好的人。 \n\n要查看完整的贡献者列表，请访问 [GitHub 贡献者页面](https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fgraphs\u002Fcontributors)。\n\n\n## 许可证\n\n本项目根据 [MIT 许可证](LICENSE) 许可。","# MTCNN 快速上手指南\n\n## 环境准备\n- **系统要求**：支持 Windows、Linux 或 macOS 系统\n- **Python 版本**：需安装 Python 3.10 或更高版本\n- **前置依赖**：无特殊依赖，安装过程将自动处理 TensorFlow >= 2.12（推荐使用国内镜像源加速下载）\n\n## 安装步骤\n使用以下命令安装 MTCNN 及其依赖（**优先推荐清华镜像源加速**）：\n\n```bash\npip install mtcnn[tensorflow] -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n```\n\n## 基本使用\n以下是最简使用示例（直接复制运行即可）：\n\n```python\nfrom mtcnn import MTCNN\nfrom mtcnn.utils.images import load_image\n\n# 创建检测器实例\ndetector = MTCNN(device=\"CPU:0\")\n\n# 加载图像\nimage = load_image(\"ivan.jpg\")\n\n# 检测图像中的人脸\nresult = detector.detect_faces(image)\n\n# 显示结果\nprint(result)\n```\n\n输出示例：\n```json\n[\n    {\n        \"box\": [277, 90, 48, 63],\n        \"keypoints\": {\n            \"nose\": [303, 131],\n            \"mouth_right\": [313, 141],\n            \"right_eye\": [314, 114],\n            \"left_eye\": [291, 117],\n            \"mouth_left\": [296, 143]\n        },\n        \"confidence\": 0.9985\n    }\n]\n```","某金融科技团队在开发移动银行App的实时人脸活体检测模块时，需精准识别用户面部特征以防止欺诈交易。\n\n### 没有 mtcnn 时\n- 检测精度不稳定：在弱光或用户戴眼镜场景下，传统Haar级联检测器漏检率超35%，导致用户反复重试活体验证。\n- 实时性能瓶颈：单帧处理耗时200ms以上，视频流卡顿明显，影响App流畅度和用户留存率。\n- 关键点缺失：无法获取眼睛、鼻尖等精确坐标，活体检测（如眨眼动作）逻辑难以实现，安全风险高。\n- 集成复杂度高：需手动编写模型加载、图像归一化和后处理代码，开发周期延长2周以上。\n- 批量处理效率低：处理用户上传的多张照片时需逐帧调用，后台任务耗时翻倍。\n\n### 使用 mtcnn 后\n- 高鲁棒性检测：在复杂光照和遮挡条件下漏检率降至5%以内，用户一次通过率提升至92%。\n- 优化实时性能：TensorFlow优化后单帧处理压缩至40ms，视频流稳定60fps，交互体验显著改善。\n- 内置关键点输出：直接提供5个面部关键点坐标，10分钟内集成眨眼活体检测逻辑，欺诈拦截率提高40%。\n- 开箱即用集成：pip安装后3行代码完成检测，开发时间缩短70%，团队快速迭代新功能。\n- 高效批量处理：单次调用支持多图输入，用户照片审核速度提升3倍，后台资源消耗降低50%。\n\nmtcnn以高精度、易集成和实时性能，为金融级人脸认证系统提供了可靠且高效的技术基石。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fipazc_mtcnn_87a279b9.jpg","ipazc","Iván de Paz Centeno","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fipazc_1bb8976d.png","PhD in Artificial Intelligence\r\nFounder at @relev-ai ","@relev-ai ","León, Spain",null,"https:\u002F\u002Fipazc.com","https:\u002F\u002Fgithub.com\u002Fipazc",[85,89],{"name":86,"color":87,"percentage":88},"Jupyter Notebook","#DA5B0B",98.2,{"name":90,"color":91,"percentage":92},"Python","#3572A5",1.8,2470,533,"2026-03-31T10:38:53","MIT",1,"未说明",{"notes":100,"python":101,"dependencies":102},"需安装TensorFlow >= 2.12；首次运行时自动下载预训练模型权重；支持通过device参数指定CPU或GPU设备","3.10+",[103],"tensorflow>=2.12",[13,14,51],[67,106,107,108,109,110,111,112],"face","detection","tensorflow","pip","package","python3","landmark","2026-03-27T02:49:30.150509","2026-04-06T05:18:04.659552",[116,121,126,131,136,141],{"id":117,"question_zh":118,"answer_zh":119,"source_url":120},4382,"如何禁用 MTCNN 的进度条输出？","对于较新版本的 Keras，使用 `import keras` 后调用 `keras.utils.disable_interactive_logging()` 可禁用进度条。如果无效，尝试移除 tqdm 模块（通过 `pip uninstall tqdm`），因为进度条可能源自该库。例如：\n```python\nimport keras\nkeras.utils.disable_interactive_logging()\n```","https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fissues\u002F121",{"id":122,"question_zh":123,"answer_zh":124,"source_url":125},4383,"如何解决 TensorFlow API 弃用警告问题？","调整导入语句为 `import tensorflow.compat.v1 as tf` 并禁用 TensorFlow v2 行为。具体步骤：在代码开头添加 `tf.disable_v2_behavior()`，并确保使用兼容模式。参考 Stack Overflow 解决方案：https:\u002F\u002Fstackoverflow.com\u002Fa\u002F56820328\u002F6825288","https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fissues\u002F40",{"id":127,"question_zh":128,"answer_zh":129,"source_url":130},4384,"MTCNN 是否支持批量处理图像以提高检测速度？","目前 MTCNN 不支持批量处理图像。开发者在评论中确认此功能尚未实现，但计划添加。建议单张图像处理，并在循环外部初始化 detector 实例以避免性能问题。例如：\n```python\nface_detector = MTCNN()  # 在循环外初始化\nfor image in images:\n    faces = face_detector.detect_faces(image)\n```","https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fissues\u002F9",{"id":132,"question_zh":133,"answer_zh":134,"source_url":135},4385,"如何防止 MTCNN 日志输出以加快大规模图像处理速度？","日志输出源自底层库（如 Keras 或 tqdm）。解决方案包括：使用 `keras.utils.disable_interactive_logging()` 禁用交互式日志，或移除 tqdm 模块。添加 `verbose=0` 参数可能无效，需直接处理依赖库。","https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fissues\u002F128",{"id":137,"question_zh":138,"answer_zh":139,"source_url":140},4386,"处理大量图像时出现内存泄漏（RAM 使用率达 100%）怎么办？","确保 MTCNN detector 实例在图像处理循环外部初始化，避免重复创建。例如：\n```python\nface_detector = MTCNN()  # 循环外初始化\nfor i in range(100000):\n    img = load_image(i)\n    faces = face_detector.detect_faces(img)\n```\n若问题 persist，检查 TensorFlow 会话管理或升级 TensorFlow 版本。","https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fissues\u002F87",{"id":142,"question_zh":143,"answer_zh":144,"source_url":145},4387,"使用 TensorFlow GPU 时出现 'MTCNN object has no attribute '_MTCNN__session'' 错误如何解决？","此错误通常由 TensorFlow 版本兼容性引起。尝试使用兼容版本（如 TensorFlow 1.x），或检查 GPU 驱动配置。若无法解决，推荐替代方案：使用 RetinaFace（PyTorch 版：https:\u002F\u002Fgithub.com\u002Fbiubug6\u002FPytorch_Retinaface；MXNet 版：https:\u002F\u002Fgithub.com\u002Fdeepinsight\u002Finsightface\u002Ftree\u002Fmaster\u002Fdetection）。","https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fissues\u002F38",[147],{"id":148,"version":149,"summary_zh":150,"released_at":151},33449,"v1.0.0","## What's Changed\r\n* Create min_face_size property by @xolott in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F16\r\n* Fix some warning messages by @akofman in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F19\r\n* new version of numpy has allow_pickle=False by default by @elmahyai in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F32\r\n* Compatibility for tensorflow 2.0 by @r-or in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F42\r\n* Fix with new numpy by @xellDart in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F34\r\n* Convert images to RGB in examples by @statsmaths in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F45\r\n* Changed the colormap to BGR during imwrite. by @raviam in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F51\r\n* Update layer_factory.py to be compatible with tf 2.0 by @xanjay in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F61\r\n* Fixed getting negative image coordinates from 'box' key. Fixes #11 by @MattyB95 in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F70\r\n* Update network\u002Ffactory.py for tf2.0 compatibility by @christian-rncl in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F85\r\n* Now you can use it as a library by @CyrusOfEden in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F86\r\n* Fixes width and height bug from negative image coordinates. Fixes #92 by @MattyB95 in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F94\r\n* Jupyter notebook example by @huberemanuel in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F106\r\n* Refactored MTCNN codebase with significant optimizations. Version 1.0.0 by @ipazc in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F133\r\n\r\n## New Contributors\r\n* @xolott made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F16\r\n* @akofman made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F19\r\n* @elmahyai made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F32\r\n* @r-or made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F42\r\n* @xellDart made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F34\r\n* @statsmaths made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F45\r\n* @raviam made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F51\r\n* @xanjay made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F61\r\n* @MattyB95 made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F70\r\n* @christian-rncl made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F85\r\n* @CyrusOfEden made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F86\r\n* @huberemanuel made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F106\r\n* @ipazc made their first contribution in https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fpull\u002F133\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fipazc\u002Fmtcnn\u002Fcommits\u002Fv1.0.0","2024-10-08T01:37:09"]