{"id":57999,"url":"https://github.com/1044197988/Awesome-Tensorflow2","name":"Awesome-Tensorflow2","description":"基于Tensorflow2开发的优秀扩展包及项目","projects_count":105,"last_synced_at":"2026-08-24T06:00:28.142Z","repository":{"id":99947742,"uuid":"211229689","full_name":"1044197988/Awesome-Tensorflow2","owner":"1044197988","description":"基于Tensorflow2开发的优秀扩展包及项目","archived":false,"fork":false,"pushed_at":"2020-02-27T14:56:00.000Z","size":133,"stargazers_count":46,"open_issues_count":0,"forks_count":23,"subscribers_count":6,"default_branch":"master","last_synced_at":"2026-07-16T12:04:51.901Z","etag":null,"topics":["automl","awesome","awesome-list","classification","detection-tensorflow2","gan","keras","nlp","segmentation","tensorflow","tensorflow-examples","tensorflow-experiments","tensorflow-models","tensorflow2","tensorflow2-awesome","tf2","tutorials"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/1044197988.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null}},"created_at":"2019-09-27T03:35:22.000Z","updated_at":"2025-12-09T13:34:14.000Z","dependencies_parsed_at":"2023-05-11T05:45:40.495Z","dependency_job_id":null,"html_url":"https://github.com/1044197988/Awesome-Tensorflow2","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/1044197988/Awesome-Tensorflow2","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/1044197988%2FAwesome-Tensorflow2","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/1044197988%2FAwesome-Tensorflow2/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/1044197988%2FAwesome-Tensorflow2/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/1044197988%2FAwesome-Tensorflow2/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/1044197988","download_url":"https://codeload.github.com/1044197988/Awesome-Tensorflow2/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/1044197988%2FAwesome-Tensorflow2/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":36283782,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-07-20T02:08:10.276Z","status":"online","status_checked_at":"2026-08-04T02:00:06.901Z","response_time":57,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"created_at":"2024-04-08T00:00:26.528Z","updated_at":"2026-08-24T06:00:28.142Z","primary_language":"Python","list_of_lists":false,"displayable":true,"categories":["Tutorials 💛💛💛💛💛\u003ca name=\"github-tutorials\" /\u003e","Model 💛💛💛💛💛\u003ca name=\"Model\" /\u003e","Other 💛💛💛💛💛\u003ca name=\"Other\" /\u003e","Projects 💛💛💛💛💛\u003ca name=\"Projects\" /\u003e"],"sub_categories":["Classification 分类","Object detection 目标检测","NLP Model 自然语言处理模型","脚本","迭代矩阵平方根归一化网络（称为快速MPN-COV），该网络非常有效，适合大规模数据集","TF2项目模板","大型基于TF2的封装扩展库","官方数据集包","AutoML","解释性工具","强化学习","Super resolution 超分辨率","TensowFlow2.0上的CBAM（卷积块注意模块）实现","Point Model 点云模型","NLP","元学习","Segmentation 分割","强大的扩展","Spektral是一个基于Keras API和TensorFlow 2的用于图深度学习的Python库。该项目的主要目标是提供一个简单而灵活的框架来创建图神经网络（GNN）。","新型数据处理技术","神经结构化学习","推荐系统和CTR预测模型","视觉较大项目","分割","目标检测","验证码识别","文字检测、识别","人脸识别","用于分布式培训，评估，模型选择和快速原型制作","TF-GAN是用于培训和评估生成对抗网络GAN的轻量级库","GCN图神经网络","RBF径向基","新型的高性能可解释的深表格式数据学习网络TabNet","优化器","用于单通道语音分离的双路径RNN","使用tensorflow 2构建的多任务学习包"],"readme":"# Awesome-Tensorflow2 💛 [![Awesome](https://awesome.re/badge.svg)](https://awesome.re)\n基于Tensorflow2开发的优秀扩展包及项目\n![Tensorflow](https://github.com/1044197988/Awesome-Tensorflow2/blob/master/Logo/Logo.jpg)\n\n# Tensorflow2说明\n2019 谷歌开发者大会于 9 月 10 日和 11 日在上海举办，大会将分享众多开发经验与工具。在第一天的 KeyNote 中，谷歌发布了很多开发工具新特性，并介绍而它们是如何构建更好的应用。值得注意的是，TensorFlow 刚刚发布了 2.0 RC01 版和 1.15，谷歌表示 1.15 是 1.x 的最后一次更新了。TensorFlow 2.0 相信大家已经非常熟悉了，它重点还是放在优化 Keras 和 Eager Execution 的能力，它希望通过这两种 API 简化整个开发流程。所以，未来的趋势肯定是Tensorflow2.0，在此我整合了许多的优秀库及项目到这个贡献库里。\n\n# 提示\n有些项目目前仍在进行中，将在未来支持Tensorflow2，这样的项目也包含在下面列表中。（意味着目前并不支持二版本）\n\n# Contents \u003ca name=\"TOC\" /\u003e👈\n\u003c!-- MarkdownTOC depth=4 --\u003e\n* [Tutorials](#github-tutorials)\n* [Model](#Model)\n* [Projects](#Projects)\n* [Other](#other)\n\u003c!-- /MarkdownTOC --\u003e \n\n## Tutorials 💛💛💛💛💛\u003ca name=\"github-tutorials\" /\u003e\n* [tensorflow.google.cn](https://tensorflow.google.cn/beta)\n* [dragen1860/TensorFlow-2.x-Tutorials](https://github.com/dragen1860/TensorFlow-2.x-Tutorials)\n* [dragen1860/Deep-Learning-with-TensorFlow-book](https://github.com/dragen1860/Deep-Learning-with-TensorFlow-book)\n* [czy36mengfei/tensorflow2_tutorials_chinese](https://github.com/czy36mengfei/tensorflow2_tutorials_chinese)\n* [yusugomori/deeplearning-tf2](https://github.com/yusugomori/deeplearning-tf2)\n* [YunYang1994/TensorFlow2.0-Examples](https://github.com/YunYang1994/TensorFlow2.0-Examples)\n\n## Model 💛💛💛💛💛\u003ca name=\"Model\" /\u003e\n### Classification 分类\n* [tensorflow/models](https://github.com/tensorflow/models)\u003cbr\u003e\n该存储库包含在TensorFlow中实现的许多不同模型。\u003cbr\u003e\n* [1044197988/TF.Keras-Commonly-used-models](https://github.com/1044197988/TF.Keras-Commonly-used-models)\u003cbr\u003e\n该贡献库为我整理的一些常用的分类、分割模型，包含分割的一些指标、损失函数，但不提供预训练模型的载入。分割模型列表如下：\u003cbr\u003e\n![Segmentation](https://github.com/1044197988/Awesome-Tensorflow2/blob/master/Logo/2.png)\n\n#### Large library 大型库\n* [qubvel/classification_models](https://github.com/qubvel/classification_models#architectures)\n\n#### Model 模型\n* [qubvel/efficientnet](https://github.com/qubvel/efficientnet)\n* [nsarang/MnasNet](https://github.com/nsarang/MnasNet)\n* [calmisential/MobileNetV3_TensorFlow2](https://github.com/calmisential/MobileNetV3_TensorFlow2)\n* [calmisential/TensorFlow2.0_ResNet](https://github.com/calmisential/TensorFlow2.0_ResNet)\n* [calmisential/TensorFlow2.0_InceptionV3](https://github.com/calmisential/TensorFlow2.0_InceptionV3)\n* [calmisential/TensorFlow2.0_Image_Classification](https://github.com/calmisential/TensorFlow2.0_Image_Classification)\n\n##### Capsnet model 胶囊网络模型\n* [prabhuomkar/hicr-capsnet](https://github.com/prabhuomkar/hicr-capsnet)\n\n##### Semi-supervised learning 半监督学习\n* [ntozer/mixmatch-tensorflow2.0](https://github.com/ntozer/mixmatch-tensorflow2.0)\n* [schatty/prototypical-networks-tf](https://github.com/schatty/prototypical-networks-tf)\n\n##### Generative-models and Self encoder 生成模型和自编码\n* [timsainb/tensorflow2-generative-models](https://github.com/timsainb/tensorflow2-generative-models)\n* [DequanZhu/GANs-collections-tf2.0_keras-eager_mode](https://github.com/DequanZhu/GANs-collections-tf2.0_keras-eager_mode)\n* [Hourout/GAN-keras](https://github.com/Hourout/GAN-keras)\n* [Net-Mist/style-transfer-tf2](https://github.com/Net-Mist/style-transfer-tf2)\n* [mnicnc404/CartoonGan-tensorflow](https://github.com/mnicnc404/CartoonGan-tensorflow)\n* [ialhashim/StyleGAN-Tensorflow2](https://github.com/ialhashim/StyleGAN-Tensorflow2)\n* [leafinity/SAGAN-tensorflow2.0](https://github.com/leafinity/SAGAN-tensorflow2.0)\n* [Lornatang/TensorFlow2-GAN](https://github.com/Lornatang/TensorFlow2-GAN)\n* [ppooiiuuyh/SinGAN-tensorflow2.0](https://github.com/ppooiiuuyh/SinGAN-tensorflow2.0)\n* [hollygrimm/tf2-cyclegan](https://github.com/hollygrimm/tf2-cyclegan)\n* [drewszurko/tensorflow-WGAN-GP](https://github.com/drewszurko/tensorflow-WGAN-GP)\n* [mgmk2/StyleGAN](https://github.com/mgmk2/StyleGAN)\n* [LynnHo/CycleGAN-Tensorflow-2](https://github.com/LynnHo/CycleGAN-Tensorflow-2)\n\n### Segmentation 分割\n#### Large library 大型库\n* [qubvel/segmentation_models](https://github.com/qubvel/segmentation_models)\n\n#### Model 模型 \n* [srihari-humbarwadi/FastFCN_TF2.0](https://github.com/srihari-humbarwadi/FastFCN_TF2.0)\n* [bonlime/keras-deeplab-v3-plus](https://github.com/bonlime/keras-deeplab-v3-plus)\n* [matterport/Mask_RCNN](https://github.com/matterport/Mask_RCNN)\n* [srihari-humbarwadi/DeepLabV3_Plus-Tensorflow2.0](https://github.com/srihari-humbarwadi/DeepLabV3_Plus-Tensorflow2.0)\n\n### Super resolution 超分辨率\n#### Model 模型\n* [krasserm/super-resolution](https://github.com/krasserm/super-resolution)\u003cbr\u003e\n包含以下模型：\u003cbr\u003e\n用于单图像超分辨率（EDSR）的增强型深度残留网络，是NTIRE 2017超分辨率挑战赛的冠军。\u003cbr\u003e\n广泛激活以实现高效，准确的图像超分辨率（WDSR），是NTIRE 2018超分辨率挑战赛（真实轨道）的获胜者。\u003cbr\u003e\n使用生成对抗网络（SRGAN）的逼真的单图像超分辨率。\u003cbr\u003e\n* [HasnainRaz/Fast-SRGAN](https://github.com/HasnainRaz/Fast-SRGAN)\n* [gs18113/ESPCN-TensorFlow2](https://github.com/gs18113/ESPCN-TensorFlow2)\n### Object detection 目标检测\n#### Model 模型\n* [zzh8829/yolov3-tf2](https://github.com/zzh8829/yolov3-tf2)\n* [srihari-humbarwadi/YOLOv1-TensorFlow2.0](https://github.com/srihari-humbarwadi/YOLOv1-TensorFlow2.0)\n* [cjpurackal/m2det-tf](https://github.com/cjpurackal/m2det-tf)\n* [reliefs/mtcnn-tensorflow2](https://github.com/reliefs/mtcnn-tensorflow2)\n* [1044197988/Centernet-Tensorflow2.0](https://github.com/1044197988/Centernet-Tensorflow2.0)\n* [calmisential/TensorFlow2.0_FasterRCNN](https://github.com/calmisential/TensorFlow2.0_FasterRCNN)\n* [Stick-To/Object-Detection-Tensorflow2](https://github.com/Stick-To/Object-Detection-Tensorflow2)\n* [hux999/tf-fcos](https://github.com/hux999/tf-fcos)\n\n### NLP Model 自然语言处理模型\n#### Large library 大型库\n* [tensorflow/tensor2tensor](https://github.com/tensorflow/tensor2tensor)\u003cbr\u003e\nTensor2Tensor或简称T2T，是一个深度学习模型和数据集的库，旨在使深度学习更易于访问并加速ML研究。Google Brain团队和用户社区的研究人员和工程师积极使用和维护T2T 。\u003cbr\u003e\n* [huggingface/transformers](https://github.com/huggingface/transformers)\u003cbr\u003e\nTensorFlow 2.0和PyTorch的最新自然语言处理，（以前称为pytorch-transformers和pytorch-pretrained-bert）提供用于自然语言理解（NLU）和自然语言生成（NLG）的最新通用架构（BERT，GPT-2，RoBERTa，XLM，DistilBert，XLNet ...） ）包含超过32种以100多种语言编写的预训练模型，以及TensorFlow 2.0和PyTorch之间的深层互操作性。\u003cbr\u003e\n\n#### Model 模型\n* [codertimo/BERT-tf2](https://github.com/codertimo/BERT-tf2)\n* [ShaneTian/TextCNN](https://github.com/ShaneTian/TextCNN)\n* [strutive07/transformer-tensorflow2.0](https://github.com/strutive07/transformer-tensorflow2.0)\n* [thisisiron/nmt-attention-tf](https://github.com/thisisiron/nmt-attention-tf)\n* [kpe/bert-for-tf2](https://github.com/kpe/bert-for-tf2)\n* [akanyaani/gpt-2-tensorflow2.0](https://github.com/akanyaani/gpt-2-tensorflow2.0)\n\n### Point Model 点云模型\n* [dgriffiths3/pointnet2-tensorflow2](https://github.com/dgriffiths3/pointnet2-tensorflow2)\n\n## Projects 💛💛💛💛💛\u003ca name=\"Projects\" /\u003e\n### 视觉较大项目\n* [giovgiac/neptune](https://github.com/giovgiac/neptune)\n### 分割\n* [Shathe/Semantic-Segmentation-Tensorflow-2](https://github.com/Shathe/Semantic-Segmentation-Tensorflow-2)\n* [1044197988/Semantic-segmentation-of-remote-sensing-images](https://github.com/1044197988/Semantic-segmentation-of-remote-sensing-images)\n### NLP\n* [jason9693/MusicTransformer-tensorflow2.0](https://github.com/jason9693/MusicTransformer-tensorflow2.0)\n* [xingchensong/Speech-Transformer-tf2.0](https://github.com/xingchensong/Speech-Transformer-tf2.0)\n* [drukka/command-words-recognition-keras](https://github.com/drukka/command-words-recognition-keras)\n* [bryanlimy/tf2-transformer-chatbot](https://github.com/bryanlimy/tf2-transformer-chatbot)\n* [gibrano/chatbot](https://github.com/gibrano/chatbot)\n### 目标检测\n* [liushuan/YOLO-V3-Tensorflow2.0-Face-Detect-via-Wider-Face](https://github.com/liushuan/YOLO-V3-Tensorflow2.0-Face-Detect-via-Wider-Face)\n* [burnpiro/tiny-face-detection-tensorflow2](https://github.com/burnpiro/tiny-face-detection-tensorflow2)\n### 验证码识别\n* [ybsdegit/captcha_keras](https://github.com/ybsdegit/captcha_keras)\n### 文字检测、识别\n* [RaidasGrisk/tf2-fots](https://github.com/RaidasGrisk/tf2-fots)\n* [RaidasGrisk/tf2-crnn](https://github.com/RaidasGrisk/tf2-crnn)\n### 人脸识别\n* [DequanZhu/FaceNet-and-FaceLoss-collections-tensorflow2.0](https://github.com/DequanZhu/FaceNet-and-FaceLoss-collections-tensorflow2.0)\n* [Fei-Wang/insightface](https://github.com/Fei-Wang/insightface)\n* [610265158/Peppa_Pig_Face_Engine](https://github.com/610265158/Peppa_Pig_Face_Engine)\n* [610265158/faceboxes-tensorflow](https://github.com/610265158/faceboxes-tensorflow)\n\n\n## Other 💛💛💛💛💛\u003ca name=\"Other\" /\u003e\n### 脚本\n* [hacker-lin/train_test_valid_split](https://github.com/hacker-lin/train_test_valid_split)\n* [ravindrabharathi/tf_utils](https://github.com/ravindrabharathi/tf_utils)\n### 解释性工具\n* [sicara/tf-explain](https://github.com/sicara/tf-explain)\n### 推荐系统和CTR预测模型\n* [JianzhouZhan/Awesome-RecSystem-Models](https://github.com/JianzhouZhan/Awesome-RecSystem-Models)\n* [cheungdaven/DeepRec](https://github.com/cheungdaven/DeepRec)\n* [Hourout/CTR-keras](https://github.com/Hourout/CTR-keras)\n* [hojinYang/recsys-implementation.tensorflow2](https://github.com/hojinYang/recsys-implementation.tensorflow2)\n* [SSSxCCC/Recommender-System](https://github.com/SSSxCCC/Recommender-System)\n### 用于分布式培训，评估，模型选择和快速原型制作\n* [zurutech/ashpy](https://github.com/zurutech/ashpy)\n### AutoML\n* [keras-team/keras-tuner](https://github.com/keras-team/keras-tuner)\n* [keras-team/autokeras](https://github.com/keras-team/autokeras)\n* [tensorflow/adanet](https://github.com/tensorflow/adanet)\n### 元学习\n* [siavash-khodadadeh/MetaLearning-TF2.0](https://github.com/siavash-khodadadeh/MetaLearning-TF2.0)\n### 强化学习\n* [danaugrs/huskarl](https://github.com/danaugrs/huskarl)\n* [keiohta/tf2rl](https://github.com/keiohta/tf2rl)\n### 神经结构化学习\n* [neural-structured-learning](https://github.com/tensorflow/neural-structured-learning)\n### 强大的扩展\n* [tensorflow/addons](https://github.com/tensorflow/addons)\n### 迭代矩阵平方根归一化网络（称为快速MPN-COV），该网络非常有效，适合大规模数据集\n* [XuChunqiao/Tensorflow-Fast-MPNCOV](https://github.com/XuChunqiao/Tensorflow-Fast-MPNCOV)\n### TF-GAN是用于培训和评估生成对抗网络GAN的轻量级库\n* [tensorflow/gan](https://github.com/tensorflow/gan)\n### 官方数据集包\n* [tensorflow/datasets](https://github.com/tensorflow/datasets)\n### NLP\n* [tensorflow/text](https://github.com/tensorflow/text)\n### 大型基于TF2的封装扩展库\n* [tensorlayer/tensorlayer](https://github.com/tensorlayer/tensorlayer)\n### TF2项目模板\n* [michaeltinsley/TF2-Project-Template](https://github.com/michaeltinsley/TF2-Project-Template)\n### GCN图神经网络\n* [breadbread1984/GCN-tf2.0](https://github.com/breadbread1984/GCN-tf2.0)\n### RBF径向基\n* [PetraVidnerova/rbf_for_tf2](https://github.com/PetraVidnerova/rbf_for_tf2)\n### 新型的高性能可解释的深表格式数据学习网络TabNet\n* [titu1994/tf-TabNet](https://github.com/titu1994/tf-TabNet)\n### 优化器\n* [OverLordGoldDragon/keras-adamw](https://github.com/OverLordGoldDragon/keras-adamw)\n### 用于单通道语音分离的双路径RNN\n* [sp-uhh/dual-path-rnn](https://github.com/sp-uhh/dual-path-rnn)\n### 新型数据处理技术\n* [AakashKumarNain/AugMix_TF2](https://github.com/AakashKumarNain/AugMix_TF2)\n### 使用tensorflow 2构建的多任务学习包\n* [AmazaspShumik/mtlearn](https://github.com/AmazaspShumik/mtlearn)\n### TensowFlow2.0上的CBAM（卷积块注意模块）实现\n* [zhangkaifang/CBAM-TensorFlow2.0](https://github.com/zhangkaifang/CBAM-TensorFlow2.0)\n### Spektral是一个基于Keras API和TensorFlow 2的用于图深度学习的Python库。该项目的主要目标是提供一个简单而灵活的框架来创建图神经网络（GNN）。\n* [danielegrattarola/spektral](https://github.com/danielegrattarola/spektral)\n","projects_url":"https://awesome.ecosyste.ms/api/v1/lists/1044197988%2Fawesome-tensorflow2/projects"}