awesome DeepLearning
1.0.0
This project is a one-stop deep learning online encyclopedia produced by PaddlePaddle. PaddlePaddle is committed to making the innovation and application of deep learning technology simpler. For more PaddlePaddle content, please visit PaddlePaddle's official website. This project covers:
?Course category: zero-basic practice deep learning , industrial practice deep learning , special courses, Paddle paddle kit course summary materials
?Book category: Paddle paddle version of "Hand-On Deep Learning"
?Treasures: Deep Learning Questions , Interviews
?Case category: Paddle paddle industry practice example library (including smart cities: fire smoke detection, safety helmet detection; intelligent manufacturing: steel defect detection, robot grabbing; Internet: financial report identification and key field extraction, etc.
From theory to practice, from scientific research to industrial applications, all kinds of learning materials are available, aiming to help developers learn and master deep learning knowledge efficiently and quickly become AI cross-border talents.
| I hope: | I can learn: |
|---|---|
| Beginner in-depth learning | Zero-basic practice deep learning |
| Advanced deep learning | 100 Questions about Deep Learning and Deep Learning in Industrial Practice |
| Interesting deep learning | Featured Courses |
| I hope: | I can learn: |
|---|---|
| Beginner in-depth learning | Zero-basic practice deep learning |
| Advanced deep learning | In-depth learning of industrial practice, special courses |
| Practical deep learning | Paddle paddle industry practice example library, Paddle paddle product courses |
AI Studio Online Course: "Zero-Basic Practice Deep Learning" : Combining theory and code, combining practice and platform, including 20-hour video courses, created by Baidu's outstanding architect, PaddlePaddle product manager and senior R&D personnel.
"Zero-Basic Practice Deep Learning" book : The supporting books for this course are published by Tsinghua Publishing House at the end of 2020, and are sold by e-commerce companies such as JD.com/Dangdang.
For the interpretation of the Transformer series content produced by PaddlePaddle Education, you can refer to the following two platforms.
Transformer Principles and Practice Series: https://aistudio.baidu.com/aistudio/education/group/info/24683
PaddlePaddle Education official account: https://aistudio.baidu.com/aistudio/personalcenter/thirdview/908086
| field | Chapter name | Course Introduction | Notebook link |
|---|---|---|---|
| NLP | Classic pre-trained language model (Part 1) - History of development of pre-trained model | Introduction to the development history of pre-trained language models, word2vec, elmo, bert, gpt, bert, some expansions. | Notebook link |
| NLP | Classic pre-trained model (Part 1) - ELMo | A comprehensive and detailed introduction to the ELMo model structure, advantages and disadvantages, etc. | Notebook link |
| NLP | Classic pre-trained model (Part 1) - Transformer | Explain the basic principles of Transformer, including Embedding, self-attention, encoder, decoder, complexity calculation, sharing mechanism and other contents. | Notebook link |
| NLP | Classic pre-trained model (Part 2) - GPT | A comprehensive and detailed introduction to the principles of GPT, pre-training and finetune mode, GPT model structure, advantages and disadvantages, etc. | Notebook link |
| NLP | Classic pre-trained model (Part 2)-BERT | A comprehensive and detailed introduction to the basic principles of BERT, the pre-training tasks and fine tune methods, the model structure of BERT itself, the advantages and disadvantages, etc. | Notebook link |
| NLP | Natural Language Understanding of Pre-trained Models-RoBERTa | Explain the improvements in natural language comprehension of pre-trained models--RoBERTa | Notebook link |
| NLP | Natural Language Understanding of Pre-trained Models-ERNIE | Explain the improvements in natural language understanding of pre-trained models: ERNIE | Notebook link |
| NLP | Natural Language Understanding of Pre-trained Models-KBERT | Explain the improvements in natural language understanding of pre-trained models: KBERT | Notebook link |
| NLP | Natural Language Understanding of Pre-trained Models-THU-ERNIE | Explain the improvements in natural language understanding of pre-trained models: THU-ERNIE | Notebook link |
| NLP | Long sequence modeling of pre-trained models-Transformer-XL | Explain the improvement of long sequence modeling of pre-trained models: Transformer-XL | Notebook link |
| NLP | Long sequence modeling of pre-trained models-XLNet | Explain the improvements in long sequence modeling of natural language comprehension: XLNet | Notebook link |
| NLP | Long sequence modeling of pre-trained models-Longformer | Explain the improvement of long sequence modeling of pre-trained models: Longformer | Notebook link |
| Model optimization | Pre-trained model-efficient structure | Punctuation prediction based on ELECTRA | Notebook link |
| Model optimization | Pre-trained model-distillation | Pre-trained model distillation algorithm: detailed explanation of Patient-KD, DistilBERT, TinyBERT, DynaBERT models, and model distillation of TinyBERT using DynaBERT strategy | Notebook link |
| CV | Transformer-Vit, DeiT in the field of image | Explain the principles of ViT and DeiT in detail | Notebook link |
| CV | Transformer-Swin Transformer in the Image Field | Explain the principle of Swin Transformer in detail | Notebook link |
| CV | Application of Transformer Model DETR in CV Field in Object Detection Task | Explain the principle of DETR and code analysis in detail | Notebook link |
return
This project changes the MXNet code implementation in the original book "Hand-On Deep Learning" to PaddlePaddle implementation. Original book authors: Aston Zhang, Li Mu, Zachary C. Lipton, Alexander J. Smora and other community contributors, GitHub address: https://github.com/d2l-ai/d2l-zh.
This project is aimed at children's shoes that are interested in deep learning, especially those who want to use PaddlePaddle for deep learning. This project does not require you to have any background knowledge of deep learning or machine learning. You only need to understand basic mathematics and programming, such as basic linear algebra, differential and probability, and basic Python programming.
The content of deep learning questions includes the basic chapters of deep learning, advanced chapters of deep learning, deep learning application chapters, reinforcement learning chapters and interview books. For details, please refer to the Paddle Knowledge Point Document Platform.
Basics of Deep Learning
Deep Learning
Convolutional neural network
Sequence Model
Advanced Deep Learning
Deep Learning Applications
Industrial Practice
Reinforcement learning
Interview Date
return
| field | Industry Cases | source | More content |
|---|---|---|---|
| Smart Industry | Statistical monitoring of traditional instruments in the factory | Paddle paddle official | More Paddle paddle cases |
| Smart Industry | Quality inspection of lithium battery separator for new energy vehicles | Paddle paddle official | More Paddle paddle cases |
| Smart Industry | Detection of surface defects of Tianchi aluminum | Paddle paddle official | More Paddle paddle cases |
| Smart Industry | Hard helmet inspection | Paddle paddle official | More Paddle paddle cases |
| Smart City | Golf course remote sensing monitoring | Paddle paddle official | More Paddle paddle cases |
| Smart City | Snow-covered semantic segmentation | Paddle paddle official | More Paddle paddle cases |
| Smart City | Face recognition for wearing masks | Paddle paddle official | More Paddle paddle cases |
| Smart transportation | Lane line segmentation and traffic light safety inspection | Paddle paddle official | More Paddle paddle cases |
| Smart transportation | 【PaddleDetection2.0 Special Project】PP-YOLOv2 | PaddlePaddleDet | More paddleDet cases |
| Smart transportation | PaddleX helps unmanned driving (Vehicle detection and lane line segmentation based on YOLOv3) | Developer BIT Kada | More Paddle paddle cases |
| Smart transportation | eblite_marker detection | Developer TobeWell | More Paddle paddle cases |
| Smart transportation | PaddleOCR: License Plate Identification | Paddle paddle developer lonely, go in quickly | More Paddle paddle cases |
| Smart farming and forestry | Agrarian land plot identification | Paddle paddle official | More Paddle paddle cases |
| Smart farming and forestry | AI worm recognition | Paddle paddle official | More Paddle paddle cases |
| Smart farming and forest | Faster and stronger! Efficient and fast PP-YOLO practical drill | PaddlePaddleDet | More paddleDet cases |
| Smart farming and forest | PaddleX Quickly Get Started-Faster RCNN Target Detection | PaddlePaddleX | More PaddleX cases |
| Smart farming and forest | AI insect detection sharing | Developer aaaLKgo | More Paddle paddle cases |
| Smart farming and forestry | Implement forest fire monitoring based on PaddleX | Paddle paddle official | More Paddle paddle cases |
| Smart medical | Classification of common Chinese herbal medicines in medicine | Paddle paddle official | More Paddle paddle cases |
| Smart medical | Eye disease recognition | Paddle paddle official | More Paddle paddle cases |
| Smart medical | Paddle-based CT imaging segmentation | Developer Code Generator | More Paddle paddle cases |
| Smart medical | PaddleHub CT imaging analysis of pneumonia | PaddlePaddleHub | More PaddleHub cases |
| Smart medical | Baseline system for predicting transmission trends of highly pathogenic infectious diseases based on Paddle paddle PGL | Paddle paddle official | More Paddle paddle cases |
| other | Person fall detection | Developer Niki_173 | More cases for this developer |
| other | Football match action positioning | Paddle paddle official | More Paddle paddle cases |
| other | Aircraft simulation based on reinforcement learning | Paddle paddle official | More Paddle paddle cases |
| other | Implement semantic matching based on ERNIE-Gram | Paddle paddle official | More Paddle paddle cases |
| other | "NLP Check-in Camp" Practical Lesson 5: Text Sentiment Analysis | PaddlePaddleNLP | More PaddlePaddleNLP cases |
| other | "NLP Classic Project Collection" 03: Select New Year's Eve Dinner using Sentiment Analysis | PaddlePaddleNLP | More PaddlePaddleNLP cases |
| other | Classification Task: How to identify the good or bad customer emotions in customer service conversations | Developers, Chinese bbking | More Paddle paddle cases |
| other | "NLP Check-in Camp" Practical Course 3: Use pre-trained models to realize the extraction of express order information | PaddlePaddleNLP | More PaddlePaddleNLP cases |
| other | Worry about the Chinese Valentine's copywriting? PaddleHub love words are generated for you (the article contains the Chinese Valentine's Day lottery) | PaddlePaddleHub | More PaddleHub cases |
| other | PCB defect detection based on PaddleDetection | Paddle paddle official | More Paddle paddle cases |
| other | Single/multi-lens pedestrian tracking based on Baidu PaddlePaddle (Unofficial Baseline) | Developer BIT Kada | More Paddle paddle cases |
| other | PaddleLite Raspberry Pi from 0 to 1: Hardhat detection small car deployment (I) | The kang on the abyss of the developer | More Paddle paddle cases |
| other | PaddleX, PP-Yolo: teach you step by step to train, encrypt and deploy target detection models | The kang on the abyss of the developer | More Paddle paddle cases |
| other | Chinese voice recognition | Paddle paddle official | More Paddle paddle cases |
| other | PaddleHub one-click OCR Chinese recognition (ultra-lightweight 8.1M model, popular) | Paddle paddle official | More Paddle paddle cases |
| other | Old Beijing City Image Repair | PaddlePaddleGAN | More PaddleGAN cases |
| other | The Secret of Song Dynasty poets reciting poems - PaddleGAN achieves accurate lip synthesis | Paddle paddle official | More Paddle paddle cases |
| other | Verification code recognition is achieved through OCR | Paddle paddle official | More Paddle paddle cases |
| other | PaddleHub one-click OCR Chinese recognition (ultra-lightweight 8.1M model, popular) | PaddlePaddleHub | More PaddleHub cases |
| other | The entire process, understand PaddlePaddle-based image segmentation from scratch | Developer nanting03 | More Paddle paddle cases |
| other | Load forecast 0.1 | Developer gaomaosheng0 | More Paddle paddle cases |
| other | AI realizes shadow plays and inherits the disappearing art | Developer Zohar | More Paddle paddle cases |
| other | 『Deep Learning 7-day check-in camp』 Face key point detection | Developer TC.Long | More Paddle paddle cases |
| Reinforcement learning | DDPG algorithm is applied to stock quantitative trading | Developer | More Paddle paddle cases |
| Technical direction | Academic Cases | source | More content |
|---|---|---|---|
| Machine Learning | Iris classification | AIStudio official | More Paddle paddle cases |
| Feedforward neural network | Boston House Price Forecast | Developer AIStudioHelper | More Paddle paddle cases |
| Image classification | Handwritten digit recognition | AIStudio official | More Paddle paddle cases |
| Image classification | Cat and dog classification | AIStudio official | More Paddle paddle cases |
| Image classification | Application of image classification network VGG in multi-expression recognition task | Developer Jerry | More Paddle paddle cases |
| Image classification | Image Classification-ResNet | Developers are stupid | More Paddle paddle cases |
| Image classification | Use PaddlePaddle to implement image classification - SE_ResNeXt | AIStudio official | More Paddle paddle cases |
| Image classification | Deeply understand Transformer-Vit, DeiT in image classification | PaddleEdu | More Paddle paddle cases |
| Image classification | Swin Transformer | PaddleEdu | More Paddle paddle cases |
| Image classification | Small sample learning (Few-Shot Learning) | Developer DeepGeGe | More Paddle paddle cases |
| Image segmentation | Classic instance segmentation model Mask RCNN | AIStudio official | More Paddle paddle cases |
| Image segmentation | PaddleSeg_DeepLabv3+ | PaddlePaddleSeg | More Paddle paddle cases |
| Image segmentation | DeepLabV3+ implementation based on PaddlePaddle | AIStudio official | More Paddle paddle cases |
| Image detection | Advanced Deep Learning - Object Detection | AIStudio official | More Paddle paddle cases |
| Image detection | A detailed explanation of the yolov3 object detection algorithm | Developer AIStudio96069 | More Paddle paddle cases |
| Image detection | Application of Transformer Model DETR in CV Field in Object Detection Task | PaddleEdu | More Paddle paddle cases |
| Video Classification | TSN Video Classification | PaddleEdu | More Paddle paddle cases |
| Video Classification | Paddle2.1 implements the classic model of video understanding—TSM | PaddleEdu | More Paddle paddle cases |
| Video Classification | Implement video classification based on Attention and Bi-LSTM | PaddleEdu | More Paddle paddle cases |
| Video Classification | TimeSformer Real Video Understanding of Transformer Model in CV Field | PaddleEdu | More Paddle paddle cases |
| GAN | Understand the classic GAN of Generative Adversarial Networks in one article (dynamic graph, VisualDL2.0) | Developer FutureSI | More Paddle paddle cases |
| GAN | StarGAN, AttGAN, STGAN algorithm based on PaddlePaddle | AIStudio official | More Paddle paddle cases |
| OCR | Text Recognition-CRNN | Developers | More Paddle paddle cases |
| NLP | Implement 9 GLUE tasks based on ERNIE | PaddleEdu | More Paddle paddle cases |
| NLP | Application of XLNet model in NLP field in sentiment analysis | PaddleEdu | More Paddle paddle cases |
| NLP | Application of ERNIE model in reading comprehension in the field of NLP | PaddleEdu | More Paddle paddle cases |
| NLP | Application of ELECTRA in the NLP field in symbol prediction | PaddleEdu | More Paddle paddle cases |
| NLP | Application of Transformer in NLP field in machine translation | PaddleEdu | More Paddle paddle cases |
| NLP | 【Paddle match】iFlytek Questions - Chinese Question Similarity Challenge 0.9+Baseline | PaddleEdu | More Paddle paddle cases |
| NLP | Implement BERT with PaddlePaddle | AIStudio official | More Paddle paddle cases |
| Multimodal | 【Paddle CLIP】Whatever you write, he draws, a little painter who belongs to you | PaddleFleet | More Paddle paddle cases |
| Reinforcement learning | From code to paper understanding and reproducing the MADDPG algorithm (PARL) | Developer Mr. Zheng_ | More Paddle paddle cases |
| recommend | [Click-through rate estimate based on DeepFM model](https://github.com/PaddlePaddle/awesome-DeepLearning/tree/master/examples/DeepFM for CTR Prediction) | PaddleEdu | More Paddle paddle cases |
| recommend | Recommended movies based on DSSM | AIStudio official | More Paddle paddle cases |
| Knowledge distillation | SSLD distillation experiment based on CIFAR100 | PaddleClas | More Paddle paddle cases |
return
| field | Competition cases | source | introduce |
|---|---|---|---|
| Machine Learning | [Paddle game] Personal loan default forecast Baseline+ 0.607 | Developer w5688414 | DataFountain personal loan default forecast, refer to the official baseline and use paddle to improve |
| NLP | 【Paddle match】iFlytek Questions - Chinese Question Similarity Challenge 0.9+Baseline | PaddleEdu | Chinese Problem Similarity Challenge Paddle Version Baseline, using paddlenlp to complete the problem similarity assessment task through fine-tuning of pre-trained models |
| NLP | PaddleHub-based emotional recognition of netizens during the epidemic | Developer CChan | This project is a solution to netizens’ emotional recognition competition during the epidemic. PaddleHub and ERNIE were used to identify emotions in Weibo texts during the epidemic. |
| NLP | 【Paddle game】Product review opinion extraction competition baseline | Developer w5688414 | DataFountain's BERT-based product review opinion extraction competition baseline, adding optimization method |
| NLP | 【Paddle game】Screen character emotional recognition baseline-precision 0.676 | Developer w5688414 | Stories character emotional recognition baseline, using bert model |
| voice | 【Paddle game】Phone synthesis | Developer XYZ_916 | 2021 Xinwang Bank Intelligent Voice Competition baseline. As of 2021.11.17, this plan is the first in the total score and the second in the works list. |
| CV | Chinese Scene Text Recognition Challenge Baseline | Xiaodu AIStudio | The baseline project of the Chinese Scene Text Recognition Challenge, used for reference by contestants |
| CV | 【Paddle game】Handwritten font OCR recognition competition baseline | Developer Pink peach | 2021 World Artificial Intelligence Innovation Competition, Handwritten Font OCR Recognition Competition Baseline |
| CV | 2020 CCF BDCI: Remote sensing image plot segmentation baseline | Developer lxastro | 2020 CCF BDCI: Baseline model library for remote sensing image plot segmentation, including baseline model training methods and competition evaluation scripts. |
| CV | The 3rd China AI+ Innovation and Entrepreneurship Competition: The No. 1 Plan for Semi-Supervised Learning Target Positioning Competition | Developer Zhang Ya Dance | The first-place plan for the semi-supervised learning target positioning competition shared the A-list score 0.81425 and the B-list score 0.80428 |
| Data Mining | 【Padddle game】Electronic Graphic Intelligent Diagnosis Competition Baseline-0.6765 | Developer w5688414 | AIWIN ECG intelligent diagnosis competition |
Return
| product | Video courses | Study Documents |
|---|---|---|
| PaddleGAN | Generate the seven-day check-in camp for the adversarial network | |
| PaddleOCR | Explanation of OCR automatic labeling gadget, interpretation of 3.5M ultra-lightweight practical OCR model, practical OCR application and deployment | |
| PaddleClas | PaddleClas series live classes | |
| PaddleDetection | Target detection 7-day check-in camp | |
| PaddleX | Detailed explanation of PaddleX instance segmentation task, detailed explanation of PaddleX object detection task, detailed explanation of PaddleX semantic segmentation task, detailed explanation of PaddleX image classification task, PaddleX client operation guide, PaddlePaddle full process development tool PaddleX | |
| PaddleHub | Tutorial on hand to convert PaddleHub models | |
| VDL | Visual analysis tools help rapid development of AI algorithms and practical demonstration of visual and tuning of deep learning algorithms | |
| High-level API | High-level API helps you quickly get started with deep learning | |
| PaddleNLP | Natural language processing based on deep learning |
Return
Thank you very much for using this project. If you have any suggestions or comments during use, you can give us feedback on Issue or contact us by scanning the QR code below. PaddlePaddle developers are very happy to be able to help you and have more in-depth communication and technical discussions with you.
The release of this project is certified by the Apache 2.0 license.
The continuous maturity of this project cannot be separated from the contribution of all developers. If you are interested in sharing deep learning knowledge, you are very welcome to contribute to us and benefit more developers.
This project welcomes any contributions and suggestions, and most contributions require your agreement to the Participant License Agreement (CLA) to declare that you have the right and actually authorize us to use your contribution.
pip install pre-commit
pre-commit install
After adding the modified code, the modified file is code-specific. Pre-commit will automatically adjust the code format and execute it once. The subsequent commit does not need to be executed again. For details, please refer to awesome-DeepLearning Submit pull request process.
Here is a list of awesome-DeepLearning contributors: yang zhou, Niki_173, Twelveeeee, buriedms, AqourAreA, zhangjin12138, rerny, LiuCongNLP, LemonCherryFu, lutianhao
