Directions include:
1) Scene Text Detection (Scene Text Detection), which detects the position of text from scene texts such as street views, and the two documents are the detection of irregular and arbitrary-shaped text;
2) Scene Text Recognition (Scene Text Recognition), which recognizes the results of scene text detection, a total of 4 articles;
3) Handwritten Text Recognition, 2 articles;
4) Scene Text Spotting (Scene Text Spotting), 1 article, that is, the real-time ABCNet algorithm proposed by scholars from South China University of Technology and the University of Adelaide, is very attractive and has been open sourced;
5) Handwritten Text Generation (Handwritten Text Generation), in order to increase the training samples of handwritten text (it feels that it can also be used to "write homework" "manually funny"), 1 article;
6) Scene Text Synthesis (Scene Text Synthesis), in order to increase the training samples of scene text, 1 article, from Megvii Technology, UnrealText uses a rendering engine to generate realistic scene text;
7) Data augmentation of text images for training of handwriting and scene text recognition algorithms, 1 article;
8) Scene Text Editor (Scene Text Editor), to replace the text in the scene text image;
9) Reconstruction of shredded paper documents, reconstruction of documents used in the field of criminal investigation after being destroyed into pieces, 1 article;
10) Text style migration, 1 article;
11) Research on Adversarial Attacks of Scene Text Recognition, 1 article;
12) Handwriting identification, 1 article.
It is worth mentioning that 10 of the 16 articles have been open sourced or are about to be open sourced. Thanks to these developers~
For papers that have been open sourced or will soon be open sourced, the code address is also attached.
You can go to:
http://openaccess.thecvf.com/CVPR2020.py
Download these papers by topic.
If you want to download all CVPR 2020 papers, please click here:
CVPR 2020 papers are fully open for download, including the main meeting and workshop
scene text detection
Deep Relational Reasoning Graph Networks for Arbitrarily Shaped Text Detection
[1].Deep Relational Reasoning Graph Network for Arbitrary Shape Text Detection
Author| Shi-Xue Zhang, Xiaobin Zhu, Jie-Bo Hou, Chang Liu, Chun Yang, Hongfa Wang, Xu-Cheng Yin
Unit | Beijing University of Science and Technology; Beijing University of Science and Technology Joint Laboratory; Tencent Technology (Shenzhen)
Code | https://github.com/GXYM/DRRG
Remarks | CVPR 2020 Oral
Interpretation | https://blog.csdn.net/SpicyCoder/article/details/105072570
[2].ContourNet: Taking a Further Step Toward Accurate Arbitrary-Shaped Scene Text Detection
Author| Yuxin Wang, Hongtao Xie, Zheng-Jun Zha, Mengting Xing, Zilong Fu, Yongdong Zhang
Unit | University of Science and Technology of China
Code | https://github.com/wangyuxin87/ContourNet
Interpretation | https://zhuanlan.zhihu.com/p/135399747
scene text recognition
On Lexical Dependence in Scene Text Recognition
[3].On Vocabulary Reliance in Scene Text Recognition
Author | Zhaoyi Wan, Jielei Zhang, Liang Zhang, Jiebo Luo, Cong Yao
Unit | Megvii; China University of Mining and Technology; University of Rochester
[4].SCATTER: Selective Context Attentional Scene Text Recognizer
作者 | Ron Litman, Oron Anschel, Shahar Tsiper, Roee Litman, Shai Mazor, R. Manmatha
Unit | Amazon Web Services
Semantic Reasoning Networks for Precise Recognition of Scene Text
[5].Towards Accurate Scene Text Recognition With Semantic Reasoning Networks
Author | Deli Yu, Xuan Li, Chengquan Zhang, Tao Liu, Junyu Han, Jingtuo Liu, Errui Ding
Unit | National University of Science and Technology; Baidu; Chinese Academy of Sciences
Code | https://github.com/chenjun2hao/SRN.pytorch
A semantically enhanced codec framework for recognizing scene text in low-quality images (blurry, uneven lighting, incomplete characters, etc.)
[6].SEED: Semantics Enhanced Encoder-Decoder Framework for Scene Text Recognition
Author | Zhi Qiao, Yu Zhou, Dongbao Yang, Yucan Zhou, Weiping Wang
Unit | Chinese Academy of Sciences; National University of Science and Technology
Code | https://github.com/Pay20Y/SEED (coming soon )
handwritten text recognition
[7].OrigamiNet: Weakly-Supervised, Segmentation-Free, One-Step, Full Page Text Recognition by learning to unfold
Author | Mohamed Yousef, Tom E. Bishop
Unit | Intuition Machines, Inc
Code | https://github.com/IntuitionMachines/OrigamiNet
Scene Text Spotting
Real-time end-to-end scene text recognition
[8].ABCNet: Real-Time Scene Text Spotting With Adaptive Bezier-Curve Network
Author | Yuliang Liu, Hao Chen, Chunhua Shen, Tong He, Lianwen Jin, Liangwei Wang
Unit | South China University of Technology; University of Adelaide;
Code | https://github.com/Yuliang-Liu/bezier\_curve\_text\_spotting
Remarks | CVPR 2020 Oral
Interpretation | https://zhuanlan.zhihu.com/p/146276834
handwritten text generation
Semi-supervised variable-length handwritten text generation, increasing text data sets, and improving the accuracy of recognition algorithms
[9].ScrabbleGAN: Semi-Supervised Varying Length Handwritten Text Generation
作者 | Sharon Fogel, Hadar Averbuch-Elor, Sarel Cohen, Shai Mazor, Roee Litman
Units | State of Israel, Amazon Rekognition; Cornell University
Code | https://github.com/amzn/convolutional-handwriting-gan
Scene Text Synthesis
Use the rendering engine to synthesize scene text, increase training samples, and improve the accuracy of recognition algorithms
[10].UnrealText: Synthesizing Realistic Scene Text Images From the Unreal
Author | WorldShangbang Long, Cong Yao
Unit | Carnegie Mellon University; Megvii
Code | https://jyouhou.github.io/UnrealText/
Interpretation | https://zhuanlan.zhihu.com/p/137406773
Data Augmentation + Text Recognition
Image Augmentation for Handwriting and Scene Text Recognition
[11].Learn to Augment: Joint Data Augmentation and Network Optimization for Text Recognition
Author | Canjie Luo, Yuanzhi Zhu, Lianwen Jin, Yongpan Wang
Unit | South China University of Technology; Ali
Code | https://github.com/Canjie-Luo/Text-Image-Augmentation
scene text editor
[12].STEFANN: Scene Text Editor Using Font Adaptive Neural Network
作者 | Prasun Roy, Saumik Bhattacharya, Subhankar Ghosh, Umapada Pal
Unit | Indian Institute of Statistics; Indian Institute of Technology
Code | https://github.com/prasunroy/stefann
Website | https://prasunroy.github.io/stefann/
Shredded Document Reconstruction
Reconstructing documents from shredded paper for forensic and other criminal investigations
[13].Fast(er) Reconstruction of Shredded Text Documents via Self-Supervised Deep Asymmetric Metric Learning
作者 | Thiago M. Paixao, Rodrigo F. Berriel, Maria CS Boeres, Alessandro L. Koerich, Claudine Badue, Alberto F. De Souza, Thiago Oliveira-Santos
Unit | IFES, Brazil; UFES, Brazil; ETS, Canada
Text style transfer
[14].SwapText: Image Based Texts Transfer in Scenes
Author | Qiangpeng Yang, Jun Huang, Wei Lin
Unit | Ali
Scene text recognition + adversarial attack
[15].What Machines See Is Not What They Get: Fooling Scene Text Recognition Models With Adversarial Text Images
Author| Xing Xu, Jiefu Chen, Jinhui Xiao, Lianli Gao, Fumin Shen, Heng Tao Shen
Unit | University of Electronic Science and Technology of China
handwriting identification
[16].Sequential Motif Profiles and Topological Plots for Offline Signature Verification
作者 | Elias N. Zois, Evangelos Zervas, Dimitrios Tsourounis, George Economou
Unit | University of West Attica; University of Paturas
END
Article source: Original CV Jun I love computer vision @微信公司号