各种机器学习开源项目精选TOP30

桃扇骨 2022-05-31 01:30 230阅读 0赞

8800个机器学习开源项目为你精选TOP30！

授权自AI科技大本营（ID: rgznai100)  
本文共图文结合，建议阅读5分钟。  
本文为大家带来了30个广受好评的机器学习开源项目。

最近，Mybridge发布了一篇文章，对比了过去一年中机器学习领域大约8800个开源项目后，选出30个2017年度优秀的开源项目，包含机器学习开源库、数据库以及其他应用程序，这些项目差不多都是在2017年1-12月发布。Mybridge AI通过受欢迎度、参与度以及其他方面对开源项目进行评定。

对于机器学习者来说，阅读开源代码并基于代码构建自己的项目，是一个非常有效的学习方法。看看以下这些Github上平均star为3558的开源项目，你错了哪些？

在开始之前，先推荐阅读

A. 神经网络：深度学习 A-ZTM : 亲手搭建人工神经网络（推荐次数68,745 , 4.5/5 stars）

链接：[http://bit.ly/2CH1WcQ][http_bit.ly_2CH1WcQ]  
![这里写图片描述][8f34d8367b34ea80cccf112e9b3ea56c.png]

B. 用Python进行深度学习的TensorFlow的完整指南（推荐次数17,834, 4.6/5 stars）

链接：[http://bit.ly/2EatVy7][http_bit.ly_2EatVy7]

![这里写图片描述][9f441abe62796beb3adc0983fa733994.png]

接下来是Mybridge精选的Top 30的项目：

1.FastText：快速文本表示和文本分类库（Github上有11786颗星，贡献者Facebook Research

源码链接：[https://github.com/facebookresearch/MUSE][https_github.com_facebookresearch_MUSE]  
![这里写图片描述][e24608af14e888ff9c28528a68389b3f.png]

2.Deep-photo-styletransfer：“Deep Photo Style Transfer”  
这篇论文的源码和数据。（GitHub 9747颗星，论文来自于康奈尔大学的Fujun Luan）

源码链接：[https://github.com/luanfujun/deep-photo-styletransfer][https_github.com_luanfujun_deep-photo-styletransfer]  
![这里写图片描述][aa9fb2d5d9f9e589e36bd0a17555a223.png]

3.用Python和命令行来实现的最简单的面部识别API（GitHub 8672颗星，贡献者Adam Geitgey）

源码链接：[https://github.com/ageitgey/face\_recognition][https_github.com_ageitgey_face_recognition]

![这里写图片描述][4cb5f512fa77cab6d41df55a65a73298.png]

4.Magenta:利用机器智能生成音乐和美术艺术品（GitHub 8113颗星）

源码链接：[https://github.com/tensorflow/magenta][https_github.com_tensorflow_magenta]

![这里写图片描述][e7d2bbad0345fafb74202a95cf119f9b.png]

5.Sonnet：基于TensorFlow的神经网络库（GitHub 573颗星，贡献者是DeepMind的Malcolm Reynolds ）

源码链接：[https://github.com/deepmind/sonnet][https_github.com_deepmind_sonnet]  
![这里写图片描述][deb3fb7adfd4460431bc39117c48e6dd.png]

6.deeplearn.js： 一个用于Web的硬件加速机器学习库（GitHub 5462颗星，贡献者是Google Brain的Nikhil Thorat）

源码链接：[https://github.com/PAIR-code/deeplearnjs][https_github.com_PAIR-code_deeplearnjs]  
![这里写图片描述][96966c083343d5d993f9c37ecc82d408.png]

7.基于TensorFlow的快速风格迁移库（GitHub 4843颗星，贡献者是MIT的Logan Engstrom）

源码链接：[https://github.com/lengstrom/fast-style-transfer][https_github.com_lengstrom_fast-style-transfer]  
![这里写图片描述][6ab5b674be72cad1e92f2376c2ae81ec.png]

8.Pysc2: 星际争霸2学习环境（GitHub 3684颗星，贡献者是DeepMind的Timo Ewalds）

源码链接：[https://github.com/deepmind/pysc2][https_github.com_deepmind_pysc2]  
![这里写图片描述][ff18138b965ab691c4cd6ed009fa63f0.png]

9.AirSim： Microsoft AI & Research开源的基于虚幻引擎的开源模拟器，用于自动驾驶（GitHub 3861颗星，贡献者是Microsoft的Shital Shah）

源码链接：[https://github.com/Microsoft/AirSim][https_github.com_Microsoft_AirSim]  
![这里写图片描述][615e25ba525ea8b1bb027c724f48215f.png]

10.acets: 机器学习数据集的可视化工具（GitHub 3371颗星，由Google Brain贡献）

源码链接：[https://github.com/PAIR-code/facets][https_github.com_PAIR-code_facets]  
![这里写图片描述][afad964eaee155f165669a9b98924d77.png]

11.Style2Paints：用AI技术为线稿快速上色的工具（GitHub 3310颗星）

源码链接：[https://github.com/lllyasviel/style2paints][https_github.com_lllyasviel_style2paints]

![这里写图片描述][2e1485865fe55474dca094f91f1365ab.png]

12.Tensor2Tensor：一个用于广义序列-序列模型的库 - Google Research（GitHub 3087颗星，贡献者是Google Brain的Ryan Sepassi）

源码链接：[https://github.com/tensorflow/tensor2tensor][https_github.com_tensorflow_tensor2tensor]

![这里写图片描述][297b9ccddd2cbdc1c7a6d8e4ec83ecdd.png]

13.基于Pytorch实现的图片-图片转换（GitHub 2847颗星，贡献者Berkeley的Jun-Yan Zhu, Ph.D）

源码地址：[https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix][https_github.com_junyanz_pytorch-CycleGAN-and-pix2pix]

![这里写图片描述][7d85f1ccaac6a642bf54f4cb5ca1856d.png]

14.Faiss：用于密集向量的高效相似性搜索库和聚类的库（GitHub 2629颗星，贡献者Facebook Research）

源码地址：[https://github.com/facebookresearch/faiss][https_github.com_facebookresearch_faiss]  
![这里写图片描述][f251d56ea149ff3a5ce27a11f33f9824.png]

15.Fashion-minist：类似于MNIST的时尚产品数据集（GitHub 2780颗星，贡献者是Zalando Tech的Han Xiao）

源码链接：[https://github.com/zalandoresearch/fashion-mnist][https_github.com_zalandoresearch_fashion-mnist]

![这里写图片描述][74b52f0edae6024e3258a48a14ed6633.png]

16.ParlAI： 一个可用在各种公开可用的对话数据集上训练和评估AI模型的框架（GitHub 2578颗星，贡献者是Facebook 的Alexander Miller）

源码链接：[https://github.com/facebookresearch/ParlAI][https_github.com_facebookresearch_ParlAI]

![这里写图片描述][df6a63d7a707641998567ba7a606ab72.png]

17.Fairseq：Facebook AI Research的序列-序列工具包（GitHub 2571颗星）

源码链接：[https://github.com/facebookresearch/fairseq][https_github.com_facebookresearch_fairseq]

![这里写图片描述][ff453c74e68202e9ad08c0ca9bdcdabb.png]

18.Pyro：基于Python和PyTorch的深度通用概率编程（GitHub 2387颗星，贡献者Uber Engineering）

源码链接：[https://github.com/uber/pyro][https_github.com_uber_pyro]

![这里写图片描述][2758271e28fa54c7e185357db1b9e48c.png]

19.iGAN：基于GAN的交互式图像生成（GitHub 2369颗星）

源码地址：[https://github.com/junyanz/iGAN][https_github.com_junyanz_iGAN]

![这里写图片描述][41b9682e1648f7ff2270d0709a7f6d37.png]

20.Deep-image-prior：用神经网络恢复图像（GitHub 2188颗星，贡献者是Skoltech的Dmitry Ulyanov, Ph.D）

源码地址：[https://github.com/DmitryUlyanov/deep-image-prior][https_github.com_DmitryUlyanov_deep-image-prior]

![这里写图片描述][7bec44b018cbf252b321cc30ccb52d02.png]

21.人脸分类：基于 Keras CNN 模型与 OpenCV ，使用fer2013/imdb 数据集进行实时面部检测和表情/性别分类（GitHub 1967颗星）

源码地址：[https://github.com/oarriaga/face\_classification][https_github.com_oarriaga_face_classification]  
![这里写图片描述][ee40447d5eba3105a0cce9d47127ad65.png]

22.Speech-to-Text-WaveNet：使用DeepMind的WaveNet和TensorFlow进行端到端句级英语语音识别（GitHub 1961颗星，贡献者是Kakao Brain的Namju Kim）

源码地址：[https://github.com/buriburisuri/speech-to-text-wavenet][https_github.com_buriburisuri_speech-to-text-wavenet]

![这里写图片描述][fc22875c1374772d53c9c70739824303.png]

23.StarGAN: 用于多域图像-图像转化的统一生成对抗网络（GitHub 1954颗星，贡献者Korea University的Yunjey Choi）

源码地址：[https://github.com/yunjey/StarGAN][https_github.com_yunjey_StarGAN]

![这里写图片描述][9c9d78cd96386148d671fa356fa78e97.png]

24.MI-agents：Unity机器学习代理（GitHub 1658颗星，贡献者Unity3D的Arthur Juliani）

源码地址：[https://github.com/Unity-Technologies/ml-agents][https_github.com_Unity-Technologies_ml-agents]

![这里写图片描述][fc442db63628238c410596b06bd2c02e.png]

25.DeepVideoAnalytics：一个分布式可视化搜索和数据分析平台（GitHub 1494颗星，贡献者是Cornell University 的Akshay Bhat）

源码地址：[https://github.com/AKSHAYUBHAT/DeepVideoAnalytics][https_github.com_AKSHAYUBHAT_DeepVideoAnalytics]  
![这里写图片描述][a910f9d2faf8159b6db10dac5f0d88ec.png]

26.OpenNMT：Torch上的开源神经机器翻译工具包（GitHub 1490颗星）

源码地址：[https://github.com/OpenNMT/OpenNMT][https_github.com_OpenNMT_OpenNMT]

![这里写图片描述][bbb90fc7758a5731a0a37e02e0aa5147.png]

27.Pix2pixHD: 用条件GAN合成和处理2048×1024的图像（GitHub 1283颗星，贡献者是英伟达科学家 Ming-Yu Liu）

源码地址：[https://github.com/NVIDIA/pix2pixHD][https_github.com_NVIDIA_pix2pixHD]  
![这里写图片描述][d00382368771f6348ef916893e665f23.png]

28.Horovod：TensorFlow 布式 训练框架（GitHub 1188 颗星，贡献者来自Uber ）

源码地址：[https://github.com/uber/horovod][https_github.com_uber_horovod]  
![这里写图片描述][e000280cf28d797bf3a18c1d3e146253.png]

29.AI-Blocks： 一个强大而直观的所见即所得界面，可让任何人创建机器学习模型（GitHub 899颗星）

源码地址：[https://github.com/MrNothing/AI-Blocks][https_github.com_MrNothing_AI-Blocks]  
![这里写图片描述][27e5704e14eddcac245bdc85c5f4a563.png]

30.Tensorflow实现的用于语音风格转换的深度神经网络（GitHub 845颗星，贡献者是Kakao Brain AI团队的Dabi Ahn）

源码地址：[https://github.com/andabi/deep-voice-conversion][https_github.com_andabi_deep-voice-conversion]  
![这里写图片描述][952d7278320502da9f757b589e8c30f7.png]

原文链接：  
[https://medium.mybridge.co/30-amazing-machine-learning-projects-for-the-past-year-v-2018-b853b8621ac7][https_medium.mybridge.co_30-amazing-machine-learning-projects-for-the-past-year-v-2018-b853b8621ac7]

[http_bit.ly_2CH1WcQ]: http://bit.ly/2CH1WcQ
[8f34d8367b34ea80cccf112e9b3ea56c.png]: /images/20220531/a2341a01c16040a8a0152caaa71f0111.png
[http_bit.ly_2EatVy7]: http://bit.ly/2EatVy7
[9f441abe62796beb3adc0983fa733994.png]: /images/20220531/62c402a201794909897fba1901406e77.png
[https_github.com_facebookresearch_MUSE]: https://github.com/facebookresearch/MUSE
[e24608af14e888ff9c28528a68389b3f.png]: /images/20220531/026dcd5e96794fcd968466f16247f164.png
[https_github.com_luanfujun_deep-photo-styletransfer]: https://github.com/luanfujun/deep-photo-styletransfer
[aa9fb2d5d9f9e589e36bd0a17555a223.png]: /images/20220531/adc5c4017ade4010afdf360a2658717d.png
[https_github.com_ageitgey_face_recognition]: https://github.com/ageitgey/face_recognition
[4cb5f512fa77cab6d41df55a65a73298.png]: /images/20220531/85275672948341128709c841f26c1427.png
[https_github.com_tensorflow_magenta]: https://github.com/tensorflow/magenta
[e7d2bbad0345fafb74202a95cf119f9b.png]: /images/20220531/cf017a3bc04e4a8487c3b00de26b1edc.png
[https_github.com_deepmind_sonnet]: https://github.com/deepmind/sonnet
[deb3fb7adfd4460431bc39117c48e6dd.png]: /images/20220531/6a251362d9ba44f4921300f0e9c6d092.png
[https_github.com_PAIR-code_deeplearnjs]: https://github.com/PAIR-code/deeplearnjs
[96966c083343d5d993f9c37ecc82d408.png]: /images/20220531/f4a3892cf76446f396acc71492089734.png
[https_github.com_lengstrom_fast-style-transfer]: https://github.com/lengstrom/fast-style-transfer
[6ab5b674be72cad1e92f2376c2ae81ec.png]: /images/20220531/cd34acd46196498bbd25c80e9b757c05.png
[https_github.com_deepmind_pysc2]: https://github.com/deepmind/pysc2
[ff18138b965ab691c4cd6ed009fa63f0.png]: /images/20220531/ea43754fd6bf4453980d76c304c1d1ad.png
[https_github.com_Microsoft_AirSim]: https://github.com/Microsoft/AirSim
[615e25ba525ea8b1bb027c724f48215f.png]: /images/20220531/ae17222dcb464f67aa0a0d0dc64c91f0.png
[https_github.com_PAIR-code_facets]: https://github.com/PAIR-code/facets
[afad964eaee155f165669a9b98924d77.png]: /images/20220531/f3605336b7334a89831f4b1c65113fc8.png
[https_github.com_lllyasviel_style2paints]: https://github.com/lllyasviel/style2paints
[2e1485865fe55474dca094f91f1365ab.png]: /images/20220531/694c7bfd1a8c4b68a4c812aad1f32c0d.png
[https_github.com_tensorflow_tensor2tensor]: https://github.com/tensorflow/tensor2tensor
[297b9ccddd2cbdc1c7a6d8e4ec83ecdd.png]: /images/20220531/eabbf16b9b9f41689891ec9df72468f4.png
[https_github.com_junyanz_pytorch-CycleGAN-and-pix2pix]: https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
[7d85f1ccaac6a642bf54f4cb5ca1856d.png]: /images/20220531/0eb13c54e3a54ff99fb939e52bd08a85.png
[https_github.com_facebookresearch_faiss]: https://github.com/facebookresearch/faiss
[f251d56ea149ff3a5ce27a11f33f9824.png]: /images/20220531/133461783ff1497da6354f150c048113.png
[https_github.com_zalandoresearch_fashion-mnist]: https://github.com/zalandoresearch/fashion-mnist
[74b52f0edae6024e3258a48a14ed6633.png]: /images/20220531/ce82b3b4f65440229f8168711fa69240.png
[https_github.com_facebookresearch_ParlAI]: https://github.com/facebookresearch/ParlAI
[df6a63d7a707641998567ba7a606ab72.png]: /images/20220531/43f208f9a10b428f94a9c2b04cfc2d79.png
[https_github.com_facebookresearch_fairseq]: https://github.com/facebookresearch/fairseq
[ff453c74e68202e9ad08c0ca9bdcdabb.png]: /images/20220531/6bf35cf8c6fb42359acaef7bad0a50b0.png
[https_github.com_uber_pyro]: https://github.com/uber/pyro
[2758271e28fa54c7e185357db1b9e48c.png]: /images/20220531/8ef8737289ed471e92e678d6f1243074.png
[https_github.com_junyanz_iGAN]: https://github.com/junyanz/iGAN
[41b9682e1648f7ff2270d0709a7f6d37.png]: /images/20220531/23cd24c4766c4544b67e56cbefa47075.png
[https_github.com_DmitryUlyanov_deep-image-prior]: https://github.com/DmitryUlyanov/deep-image-prior
[7bec44b018cbf252b321cc30ccb52d02.png]: /images/20220531/475b9b83c9ed42468f0b6efdfbaaa642.png
[https_github.com_oarriaga_face_classification]: https://github.com/oarriaga/face_classification
[ee40447d5eba3105a0cce9d47127ad65.png]: /images/20220531/fe147f0b7ad1442dbd75d7da8144d0de.png
[https_github.com_buriburisuri_speech-to-text-wavenet]: https://github.com/buriburisuri/speech-to-text-wavenet
[fc22875c1374772d53c9c70739824303.png]: /images/20220531/c81511ef57564a6aac2cf02a35df8c61.png
[https_github.com_yunjey_StarGAN]: https://github.com/yunjey/StarGAN
[9c9d78cd96386148d671fa356fa78e97.png]: /images/20220531/5c1179058ec843308db8a3d13205b68d.png
[https_github.com_Unity-Technologies_ml-agents]: https://github.com/Unity-Technologies/ml-agents
[fc442db63628238c410596b06bd2c02e.png]: /images/20220531/bb55d12b78ea46358bc9c93bba7b856f.png
[https_github.com_AKSHAYUBHAT_DeepVideoAnalytics]: https://github.com/AKSHAYUBHAT/DeepVideoAnalytics
[a910f9d2faf8159b6db10dac5f0d88ec.png]: /images/20220531/640f7b4d48eb4d7eb2359319af48030a.png
[https_github.com_OpenNMT_OpenNMT]: https://github.com/OpenNMT/OpenNMT
[bbb90fc7758a5731a0a37e02e0aa5147.png]: /images/20220531/ec64524f640f4c50824a0fa4a61ca726.png
[https_github.com_NVIDIA_pix2pixHD]: https://github.com/NVIDIA/pix2pixHD
[d00382368771f6348ef916893e665f23.png]: /images/20220531/3435e063b12e44c8a8322137d7058fcc.png
[https_github.com_uber_horovod]: https://github.com/uber/horovod
[e000280cf28d797bf3a18c1d3e146253.png]: /images/20220531/9b9a079c247944ce98fc531b9f36438f.png
[https_github.com_MrNothing_AI-Blocks]: https://github.com/MrNothing/AI-Blocks
[27e5704e14eddcac245bdc85c5f4a563.png]: /images/20220531/898ed3d66c81493eb80e803bf1e64e01.png
[https_github.com_andabi_deep-voice-conversion]: https://github.com/andabi/deep-voice-conversion
[952d7278320502da9f757b589e8c30f7.png]: /images/20220531/4523ae8c171f4f3da93545ec45f30c44.png
[https_medium.mybridge.co_30-amazing-machine-learning-projects-for-the-past-year-v-2018-b853b8621ac7]: https://medium.mybridge.co/30-amazing-machine-learning-projects-for-the-past-year-v-2018-b853b8621ac7