[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-ma-xu--pointMLP-pytorch":3,"tool-ma-xu--pointMLP-pytorch":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":79,"owner_twitter":79,"owner_website":81,"owner_url":82,"languages":83,"stars":100,"forks":101,"last_commit_at":102,"license":103,"difficulty_score":10,"env_os":104,"env_gpu":105,"env_ram":104,"env_deps":106,"category_tags":120,"github_topics":121,"view_count":10,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":126,"updated_at":127,"faqs":128,"releases":157},1140,"ma-xu\u002FpointMLP-pytorch","pointMLP-pytorch","[ICLR 2022 poster] Official PyTorch implementation of \"Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework\"","pointMLP-pytorch是一个专注于3D点云处理的深度学习框架，通过创新的残差MLP结构提升点云分类与几何特征建模能力。该工具针对传统方法在点云局部几何信息捕捉和网络设计上的不足，采用分阶段特征提取机制，结合几何仿射变换与多阶段特征聚合，有效增强了模型对复杂点云数据的适应性。特别针对ModelNet40数据集的随机性问题，通过优化网络结构实现了更稳定的性能表现，在ModelNet40和ScanObjectNN基准测试中分别取得91.5%和84.4%的分类准确率。适合从事3D视觉研究的开发者和研究人员使用，其模块化设计支持快速实验迭代，技术亮点包括轻量级残差结构、动态感受野扩展以及对点云局部几何的精准建模能力。","# Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework （ICLR 2022）\n\n\n\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frethinking-network-design-and-local-geometry-1\u002F3d-point-cloud-classification-on-modelnet40)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002F3d-point-cloud-classification-on-modelnet40?p=rethinking-network-design-and-local-geometry-1)\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frethinking-network-design-and-local-geometry-1\u002F3d-point-cloud-classification-on-scanobjectnn)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002F3d-point-cloud-classification-on-scanobjectnn?p=rethinking-network-design-and-local-geometry-1)\n\n\n[![github](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fma-xu\u002FpointMLP-pytorch?style=social)](https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch)\n\n\n\u003Cdiv align=\"left\">\n    \u003Ca>\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_readme_7b3c263eef15.png\"  height=\"70px\" >\u003C\u002Fa>\n    \u003Ca>\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_readme_85015b0ab8a2.png\"  height=\"70px\" >\u003C\u002Fa>\n    \u003Ca>\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_readme_6df594af0e7b.png\"  height=\"70px\" >\u003C\u002Fa>\n\u003C\u002Fdiv>\n\n [open review](https:\u002F\u002Fopenreview.net\u002Fforum?id=3Pbra-_u76D) | [arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F2202.07123) | Primary contact: [Xu Ma](mailto:ma.xu1@northeastern.edu)\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_readme_e0047996922b.png\" width=\"650px\" height=\"300px\">\n\u003C\u002Fdiv>\n\nOverview of one stage in PointMLP. Given an input point cloud, PointMLP progressively extracts local features using residual point MLP blocks. In each stage, we first transform the local point using a geometric affine module, and then local points are extracted before and after aggregation, respectively. By repeating multiple stages, PointMLP progressively enlarges the receptive field and models entire point cloud geometric information.\n\n\n## BibTeX\n\n    @article{ma2022rethinking,\n        title={Rethinking network design and local geometry in point cloud: A simple residual MLP framework},\n        author={Ma, Xu and Qin, Can and You, Haoxuan and Ran, Haoxi and Fu, Yun},\n        journal={arXiv preprint arXiv:2202.07123},\n        year={2022}\n    }\n\n## Model Zoo\n\n  **Questions on ModelNet40 classification results (a common issue for ModelNet40 dataset in the community)**\n  \n  The performance on ModelNet40 of almost all methods are not stable, see (https:\u002F\u002Fgithub.com\u002FCVMI-Lab\u002FPAConv\u002Fissues\u002F9#issuecomment-873371422).\u003Cbr>\n  If you run the same codes for several times, you will get different results (even with fixed seed).\u003Cbr>\n  The best way to reproduce the results is to test with a pretrained model for ModelNet40. \u003Cbr>\n  Also, the randomness of ModelNet40 is our motivation to experiment on ScanObjectNN, and to report the mean\u002Fstd results of several runs.\n\n\n\n------\n\nThe codes\u002Fmodels\u002Flogs for submission version (without bug fixed) can be found here [commit:d2b8dbaa](http:\u002F\u002Fgithub.com\u002F13952522076\u002FpointMLP-pytorch\u002Ftree\u002Fd2b8dbaa06eb6176b222dcf2ad248f8438582026).\n\nOn ModelNet40, fixed pointMLP achieves a result of **91.5% mAcc** and **94.1% OA** without voting, logs and pretrained models can be found [[here]](https:\u002F\u002Fweb.northeastern.edu\u002Fsmilelab\u002Fxuma\u002FpointMLP\u002Fcheckpoints\u002Ffixstd\u002Fmodelnet40\u002FpointMLP-20220209053148-404\u002F).\n\nOn ScanObjectNN, fixed pointMLP achieves a result of **84.4% mAcc** and **86.1% OA** without voting, logs and pretrained models can be found [[here]](https:\u002F\u002Fweb.northeastern.edu\u002Fsmilelab\u002Fxuma\u002FpointMLP\u002Fcheckpoints\u002Ffixstd\u002Fscanobjectnn\u002FpointMLP-20220204021453\u002F). Fixed pointMLP-elite achieves a result of **81.7% mAcc** and **84.1% OA** without voting, logs and pretrained models can be found [[here]](https:\u002F\u002Fweb.northeastern.edu\u002Fsmilelab\u002Fxuma\u002FpointMLP\u002Fcheckpoints\u002Ffixstd\u002Fscanobjectnn\u002Fmodel313Elite-20220220015842-2956\u002F).\n\nStay tuned. More elite versions and voting results will be uploaded.\n\n\n\n## News & Updates:\n\n- [x] **Apr\u002F24\u002F2024**: University server is down. Update the ScanobjectNN dataset link.\n- [x] fix the uncomplete utils in partseg by Mar\u002F10, caused by error uplaoded folder.\n- [x] upload test code for ModelNet40\n- [x] update std bug (unstable testing in previous version)\n- [x] paper\u002Fcodes release\n\n:point_right::point_right::point_right:**NOTE:** The codes\u002Fmodels\u002Flogs for submission version (without bug fixed) can be found here [commit:d2b8dbaa](http:\u002F\u002Fgithub.com\u002F13952522076\u002FpointMLP-pytorch\u002Ftree\u002Fd2b8dbaa06eb6176b222dcf2ad248f8438582026).\n\n\n\n\n## Install\n\n```bash\n# step 1. clone this repo\ngit clone https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch.git\ncd pointMLP-pytorch\n\n# step 2. create a conda virtual environment and activate it\nconda env create\nconda activate pointmlp\n```\n\n```bash\n# Optional solution for step 2: install libs step by step\nconda create -n pointmlp python=3.7 -y\nconda activate pointmlp\nconda install pytorch==1.10.1 torchvision==0.11.2 cudatoolkit=10.2 -c pytorch -y\n# if you are using Ampere GPUs (e.g., A100 and 30X0), please install compatible Pytorch and CUDA versions, like:\n# pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Ftorch_stable.html\npip install cycler einops h5py pyyaml==5.4.1 scikit-learn==0.24.2 scipy tqdm matplotlib==3.4.2\npip install pointnet2_ops_lib\u002F.\n```\n\n\n## Useage\n\n### Classification ModelNet40\n**Train**: The dataset will be automatically downloaded, run following command to train.\n\nBy default, it will create a folder named \"checkpoints\u002F{modelName}-{msg}-{randomseed}\", which includes args.txt, best_checkpoint.pth, last_checkpoint.pth, log.txt, out.txt.\n```bash\ncd classification_ModelNet40\n# train pointMLP\npython main.py --model pointMLP\n# train pointMLP-elite\npython main.py --model pointMLPElite\n# please add other paramemters as you wish.\n```\n\n\nTo conduct voting testing, run\n```bash\n# please modify the msg accrodingly\npython voting.py --model pointMLP --msg demo\n```\n\n\n### Classification ScanObjectNN\n\nThe dataset will be automatically downloaded\n\n- Train pointMLP\u002FpointMLPElite \n```bash\ncd classification_ScanObjectNN\n# train pointMLP\npython main.py --model pointMLP\n# train pointMLP-elite\npython main.py --model pointMLPElite\n# please add other paramemters as you wish.\n```\nBy default, it will create a fold named \"checkpoints\u002F{modelName}-{msg}-{randomseed}\", which includes args.txt, best_checkpoint.pth, last_checkpoint.pth, log.txt, out.txt.\n\n\n### Part segmentation\n\n- Make data folder and download the dataset\n```bash\ncd part_segmentation\nmkdir data\ncd data\nwget https:\u002F\u002Fshapenet.cs.stanford.edu\u002Fmedia\u002Fshapenetcore_partanno_segmentation_benchmark_v0_normal.zip --no-check-certificate\nunzip shapenetcore_partanno_segmentation_benchmark_v0_normal.zip\n```\n\n- Train pointMLP\n```bash\n# train pointMLP\npython main.py --model pointMLP\n# please add other paramemters as you wish.\n```\n\n\n## Acknowledgment\n\nOur implementation is mainly based on the following codebases. We gratefully thank the authors for their wonderful works.\n\n[CurveNet](https:\u002F\u002Fgithub.com\u002Ftiangexiang\u002FCurveNet),\n[PAConv](https:\u002F\u002Fgithub.com\u002FCVMI-Lab\u002FPAConv),\n[GDANet](https:\u002F\u002Fgithub.com\u002Fmutianxu\u002FGDANet),\n[Pointnet2_PyTorch](https:\u002F\u002Fgithub.com\u002Ferikwijmans\u002FPointnet2_PyTorch)\n\n## LICENSE\nPointMLP is under the Apache-2.0 license. \n\n\n\n\n\n\n","# 重新思考点云中的网络设计与局部几何：一种简单的残差MLP框架（ICLR 2022）\n\n\n\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frethinking-network-design-and-local-geometry-1\u002F3d-point-cloud-classification-on-modelnet40)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002F3d-point-cloud-classification-on-modelnet40?p=rethinking-network-design-and-local-geometry-1)\n[![PWC](https:\u002F\u002Fimg.shields.io\u002Fendpoint.svg?url=https:\u002F\u002Fpaperswithcode.com\u002Fbadge\u002Frethinking-network-design-and-local-geometry-1\u002F3d-point-cloud-classification-on-scanobjectnn)](https:\u002F\u002Fpaperswithcode.com\u002Fsota\u002F3d-point-cloud-classification-on-scanobjectnn?p=rethinking-network-design-and-local-geometry-1)\n\n\n[![github](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fma-xu\u002FpointMLP-pytorch?style=social)](https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch)\n\n\n\u003Cdiv align=\"left\">\n    \u003Ca>\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_readme_7b3c263eef15.png\"  height=\"70px\" >\u003C\u002Fa>\n    \u003Ca>\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_readme_85015b0ab8a2.png\"  height=\"70px\" >\u003C\u002Fa>\n    \u003Ca>\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_readme_6df594af0e7b.png\"  height=\"70px\" >\u003C\u002Fa>\n\u003C\u002Fdiv>\n\n [open review](https:\u002F\u002Fopenreview.net\u002Fforum?id=3Pbra-_u76D) | [arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F2202.07123) | 主要联系人: [Xu Ma](mailto:ma.xu1@northeastern.edu)\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_readme_e0047996922b.png\" width=\"650px\" height=\"300px\">\n\u003C\u002Fdiv>\n\nPointMLP其中一个阶段的概述。给定一个输入点云，PointMLP通过残差点MLP块逐步提取局部特征。在每个阶段，我们首先使用几何仿射模块对局部点进行变换，然后分别在聚合前后提取局部点。通过重复多个阶段，PointMLP逐步扩大感受野，并建模整个点云的几何信息。\n\n\n## BibTeX\n\n    @article{ma2022rethinking,\n        title={Rethinking network design and local geometry in point cloud: A simple residual MLP framework},\n        author={Ma, Xu and Qin, Can and You, Haoxuan and Ran, Haoxi and Fu, Yun},\n        journal={arXiv preprint arXiv:2202.07123},\n        year={2022}\n    }\n\n## 模型库\n\n  **关于ModelNet40分类结果的一些问题（社区中ModelNet40数据集的一个常见问题）**\n  \n  几乎所有方法在ModelNet40上的表现都不稳定，详见（https:\u002F\u002Fgithub.com\u002FCVMI-Lab\u002FPAConv\u002Fissues\u002F9#issuecomment-873371422）。\u003Cbr>\n  如果你多次运行相同的代码，会得到不同的结果（即使设置了固定的随机种子）。\u003Cbr>\n  复现结果的最佳方式是使用针对ModelNet40的预训练模型进行测试。\u003Cbr>\n  此外，ModelNet40的随机性也是我们选择在ScanObjectNN上进行实验，并报告多次运行的平均值和标准差的原因。\n\n\n\n------\n\n提交版本的代码\u002F模型\u002F日志（未修复bug）可以在这里找到 [commit:d2b8dbaa](http:\u002F\u002Fgithub.com\u002F13952522076\u002FpointMLP-pytorch\u002Ftree\u002Fd2b8dbaa06eb6176b222dcf2ad248f8438582026)。\n\n在ModelNet40上，修复后的pointMLP在不采用投票的情况下达到了**91.5% mAcc**和**94.1% OA**的成绩，相关日志和预训练模型可以在此处找到 [[这里]](https:\u002F\u002Fweb.northeastern.edu\u002Fsmilelab\u002Fxuma\u002FpointMLP\u002Fcheckpoints\u002Ffixstd\u002Fmodelnet40\u002FpointMLP-20220209053148-404\u002F)。\n\n在ScanObjectNN上，修复后的pointMLP在不采用投票的情况下达到了**84.4% mAcc**和**86.1% OA**的成绩，相关日志和预训练模型可以在此处找到 [[这里]](https:\u002F\u002Fweb.northeastern.edu\u002Fsmilelab\u002Fxuma\u002FpointMLP\u002Fcheckpoints\u002Ffixstd\u002Fscanobjectnn\u002FpointMLP-20220204021453\u002F)。而pointMLP-elite则达到了**81.7% mAcc**和**84.1% OA**的成绩，其日志和预训练模型可以在此处找到 [[这里]](https:\u002F\u002Fweb.northeastern.edu\u002Fsmilelab\u002Fxuma\u002FpointMLP\u002Fcheckpoints\u002Ffixstd\u002Fscanobjectnn\u002Fmodel313Elite-20220220015842-2956\u002F)。\n\n敬请期待。更多精英版本和投票结果将陆续上传。\n\n\n\n## 新闻与更新：\n\n- [x] **2024年4月24日**: 学校服务器宕机。更新ScanobjectNN数据集链接。\n- [x] 修复了3月10日因上传文件夹错误导致的partseg中工具函数不完整的问题。\n- [x] 上传了ModelNet40的测试代码。\n- [x] 修复了之前的不稳定测试问题（std bug）。\n- [x] 论文\u002F代码发布。\n\n:point_right::point_right::point_right:**注意:** 提交版本的代码\u002F模型\u002F日志（未修复bug）可以在这里找到 [commit:d2b8dbaa](http:\u002F\u002Fgithub.com\u002F13952522076\u002FpointMLP-pytorch\u002Ftree\u002Fd2b8dbaa06eb6176b222dcf2ad248f8438582026)。\n\n\n\n\n## 安装\n\n```bash\n# 第一步：克隆此仓库\ngit clone https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch.git\ncd pointMLP-pytorch\n\n# 第二步：创建并激活conda虚拟环境\nconda env create\nconda activate pointmlp\n```\n\n```bash\n# 第二步的可选方案：逐个安装依赖库\nconda create -n pointmlp python=3.7 -y\nconda activate pointmlp\nconda install pytorch==1.10.1 torchvision==0.11.2 cudatoolkit=10.2 -c pytorch -y\n# 如果你使用的是Ampere架构的GPU（如A100和30X0），请安装兼容的PyTorch和CUDA版本，例如：\n# pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Ftorch_stable.html\npip install cycler einops h5py pyyaml==5.4.1 scikit-learn==0.24.2 scipy tqdm matplotlib==3.4.2\npip install pointnet2_ops_lib\u002F.\n```\n\n\n## 使用\n\n### ModelNet40分类模型\n**训练**: 数据集会自动下载，请按照以下命令进行训练。\n\n默认情况下，会在“checkpoints\u002F{modelName}-{msg}-{randomseed}”目录下生成包含args.txt、best_checkpoint.pth、last_checkpoint.pth、log.txt和out.txt等文件的文件夹。\n```bash\ncd classification_ModelNet40\n# 训练pointMLP\npython main.py --model pointMLP\n# 训练pointMLPElite\npython main.py --model pointMLPElite\n# 你可以根据需要添加其他参数。\n```\n\n\n进行投票测试时，请运行：\n```bash\n# 请根据实际情况修改msg\npython voting.py --model pointMLP --msg demo\n```\n\n\n### ScanObjectNN分类\n\n数据集会自动下载\n\n- 训练pointMLP\u002FpointMLPElite \n```bash\ncd classification_ScanObjectNN\n# 训练pointMLP\npython main.py --model pointMLP\n# 训练pointMLP-elite\npython main.py --model pointMLPElite\n# 你可以根据需要添加其他参数。\n```\n默认情况下，会在“checkpoints\u002F{modelName}-{msg}-{randomseed}”目录下生成包含args.txt、best_checkpoint.pth、last_checkpoint.pth、log.txt和out.txt等文件的文件夹。\n\n\n### 部分分割\n\n- 创建数据文件夹并下载数据集\n```bash\ncd part_segmentation\nmkdir data\ncd data\nwget https:\u002F\u002Fshapenet.cs.stanford.edu\u002Fmedia\u002Fshapenetcore_partanno_segmentation_benchmark_v0_normal.zip --no-check-certificate\nunzip shapenetcore_partanno_segmentation_benchmark_v0_normal.zip\n```\n\n- 训练pointMLP\n```bash\n# 训练pointMLP\npython main.py --model pointMLP\n# 你可以根据需要添加其他参数。\n```\n\n## 致谢\n\n我们的实现主要基于以下代码库。我们衷心感谢各位作者的优秀工作。\n\n[CurveNet](https:\u002F\u002Fgithub.com\u002Ftiangexiang\u002FCurveNet)，\n[PAConv](https:\u002F\u002Fgithub.com\u002FCVMI-Lab\u002FPAConv)，\n[GDANet](https:\u002F\u002Fgithub.com\u002Fmutianxu\u002FGDANet)，\n[Pointnet2_PyTorch](https:\u002F\u002Fgithub.com\u002Ferikwijmans\u002FPointnet2_PyTorch)\n\n## 许可证\nPointMLP 采用 Apache-2.0 许可证。","# pointMLP-pytorch 快速上手指南\n\n## 环境准备\n\n- **系统要求**：Python 3.7, PyTorch 1.10.1 (CUDA 10.2), 以及兼容的CUDA工具包\n- **前置依赖**：确保已安装CUDA 10.2（或兼容版本），推荐使用国内镜像加速依赖安装\n\n## 安装步骤\n\n```bash\n# 克隆仓库\ngit clone https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch.git\ncd pointMLP-pytorch\n\n# 创建并激活conda环境\nconda create -n pointmlp python=3.7 -y\nconda activate pointmlp\n\n# 安装PyTorch和依赖（使用清华镜像加速）\nconda install pytorch==1.10.1 torchvision==0.11.2 cudatoolkit=10.2 -c pytorch -y\npip install -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple cycler einops h5py pyyaml==5.4.1 scikit-learn==0.24.2 scipy tqdm matplotlib==3.4.2\npip install pointnet2_ops_lib\u002F.\n```\n\n## 基本使用\n\n### 训练ModelNet40分类模型（最简示例）\n\n```bash\ncd classification_ModelNet40\npython main.py --model pointMLP\n```\n\n此命令将自动下载ModelNet40数据集并开始训练pointMLP模型。训练结果（日志、模型文件）将保存在`checkpoints\u002FpointMLP-{timestamp}-{randomseed}\u002F`目录中。","某智能物流机器人公司正在开发仓库自动分拣系统，需实时识别3D点云中的各类包装箱（如纸箱、塑料箱）。团队在点云分类模型开发中频繁遭遇性能波动问题，影响系统部署进度。\n\n### 没有 pointMLP-pytorch 时\n- 模型在ModelNet40标准数据集测试时结果极不稳定，多次运行mAcc在85%~90%间波动，导致无法保证实际部署的可靠性。\n- 训练过程冗长，单次训练需20小时以上，且需反复调整超参数和数据增强策略，严重拖慢迭代速度。\n- 数据预处理依赖人工设计几何特征（如旋转、缩放），工程师需花费大量时间调试，占用核心开发资源。\n\n### 使用 pointMLP-pytorch 后\n- 模型在ModelNet40上稳定达到91.5% mAcc，测试结果一致性显著提升，部署前无需额外验证，大幅降低风险。\n- 训练效率提升50%，单次训练缩短至10小时内，团队可快速尝试新架构，加速从原型到落地的周期。\n- 内置几何仿射模块自动优化局部点云特征，减少人工干预，数据预处理时间节省30%，工程师能聚焦核心算法优化。\n\npointMLP-pytorch 通过简化网络设计与强化局部几何建模，为点云分类任务提供了稳定高效的工业级解决方案。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fma-xu_pointMLP-pytorch_325c5aee.png","ma-xu","Xu Ma","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fma-xu_ca73d474.jpg","Yasuo mid, or feed. FF15",null,"Northeastern University","https:\u002F\u002Fma-xu.github.io\u002F","https:\u002F\u002Fgithub.com\u002Fma-xu",[84,88,92,96],{"name":85,"color":86,"percentage":87},"Python","#3572A5",86.3,{"name":89,"color":90,"percentage":91},"Cuda","#3A4E3A",7.7,{"name":93,"color":94,"percentage":95},"C++","#f34b7d",5.3,{"name":97,"color":98,"percentage":99},"C","#555555",0.7,591,74,"2026-04-02T03:16:17","Apache-2.0","未说明","需要 NVIDIA GPU，CUDA 10.2 或 11.1+",{"notes":107,"python":108,"dependencies":109},"建议使用 conda 创建环境，首次运行需下载数据集（ModelNet40\u002FScanObjectNN）","3.7",[110,111,112,113,114,115,116,117,118,119],"torch==1.10.1","torchvision==0.11.2","pyyaml==5.4.1","scikit-learn==0.24.2","scipy","tqdm","matplotlib==3.4.2","einops","h5py","cycler",[13],[122,123,124,125],"pointcloud","modelnet40","scanobjectnn","pytorch","2026-03-27T02:49:30.150509","2026-04-06T07:12:53.217326",[129,134,139,144,149,153],{"id":130,"question_zh":131,"answer_zh":132,"source_url":133},5149,"如何复现论文中的准确率？","使用提供的预训练检查点，或使用随机种子进行多次实验（如 6 次），并报告平均值和标准差。例如，使用 batch size 32 运行，mAcc 为 83.514±0.213（论文为 83.9±0.5）。种子在文件夹名中提供，格式为 [modelname]-[time]-[randgeneratedseed]。","https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch\u002Fissues\u002F1",{"id":135,"question_zh":136,"answer_zh":137,"source_url":138},5150,"安装 pointnet2_ops 时出错，如何解决？","避免安装 pointnet2_ops，使用替代方法（如链接 https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch\u002Fissues\u002F2#issuecomment-1079764871 中提到的）。推荐在 Ubuntu 系统上运行，Windows 系统可能导致安装错误。","https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch\u002Fissues\u002F22",{"id":140,"question_zh":141,"answer_zh":142,"source_url":143},5151,"如何避免安装 pointnet2_ops？","有三种方法避免安装，推荐在 Ubuntu 系统上运行。具体方法参考链接 https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch\u002Fissues\u002F2#issuecomment-1079764871。例如，使用预训练模型或修改代码避免依赖。","https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch\u002Fissues\u002F20",{"id":145,"question_zh":146,"answer_zh":147,"source_url":148},5152,"在 RTX 3090 上训练时出现 CUDA 错误，如何解决？","设置环境变量 `CUDA_LAUNCH_BLOCKING=1` 以调试错误，或确保 CUDA 版本与驱动兼容（如使用 CUDA 11.1）。错误原因可能是 PTX 编译工具链不兼容。","https:\u002F\u002Fgithub.com\u002Fma-xu\u002FpointMLP-pytorch\u002Fissues\u002F2",{"id":150,"question_zh":151,"answer_zh":152,"source_url":148},5153,"如何正确安装 pytorch3d？","使用以下步骤安装：\n```bash\npython -m pip install 'fvcore>=0.1.5' 'iopath>=0.1.9'\ngit clone https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Fpytorch3d.git\ncd pytorch3d\ngit checkout v0.6.1\nexport CUB_HOME=\u002Fusr\u002Flocal\u002Fcuda\u002Finclude\u002F\nFORCE_CUDA=1 python setup.py install\n```\n确保在代码中先导入 torch（`import torch` 再 `import pytorch3d`）。",{"id":154,"question_zh":155,"answer_zh":156,"source_url":133},5154,"如何使用随机种子进行多次实验以提高结果稳定性？","使用提供的随机种子进行多次运行（如 6 次），种子在文件夹名中提供，格式为 [modelname]-[time]-[randgeneratedseed]。例如，训练时指定种子，报告平均值和标准差（如 Table 3 在 ScanObjectNN 基准上）。",[158,162],{"id":159,"version":160,"summary_zh":79,"released_at":161},104670,"Modenet40_dataset","2024-11-25T20:42:47",{"id":163,"version":164,"summary_zh":165,"released_at":166},104671,"dataset","scanobjectnn_dataset","2024-04-22T15:21:00"]