method: Clova AI / Lens2018-09-17
Authors: Seolki Baek, Geonmo Gu, Jeongo Seo
Description: Our model is featured by CNN/RNN-based encoder and Hybrid CTC/Attention decoder. Moreover we proposed new text synthesis tools to make our model robust and high performance in the wild.
method: TencentAILab2018-04-24
Authors: Jingchao Zhou, Tianlin Gao, Zheng Zhou, Zhifeng Li
Description: We train a network to recognize the word images. First, we correct the oblique and vertical arranged text lines using tranditional OCR technologies. Second, we generate several batches of synthesized images with similar style and arrangement as training samples. Last, we adopt DenseNet as the backbone to extract features, Bi-direction LSTM to learn sequential information, and CTC as the transcription layer.
method: Tencent-OCR+2017-06-30
Authors: Chunchao Guo, Weichen Zhang, Yi Li, Hui Song, Ming Liu, Hongfa Wang, Lei Xiao
Description: Data Platform Department, Tencent. We adapt CNN-LSTM-CTC architecture to recognize the text line. In addition, a knowledge-based post processing is used for adjusting the result.
| Date | Method | Total Edit distance (case sensitive) | Correctly Recognised Words (case sensitive) | T.E.D. (case insensitive) | C.R.W. (case insensitive) | |||
|---|---|---|---|---|---|---|---|---|
| 2018-09-17 | Clova AI / Lens | 101.3546 | 94.60% | 92.7729 | 94.93% | |||
| 2018-04-24 | TencentAILab | 112.8026 | 92.34% | 101.5090 | 93.09% | |||
| 2017-06-30 | Tencent-OCR+ | 158.3418 | 89.83% | 121.7667 | 91.24% | |||
| 2017-07-01 | HIK_OCR | 198.6483 | 88.54% | 179.4190 | 89.17% | |||
| 2017-06-29 | baseline | 1,749.8736 | 34.51% | 1,501.2280 | 42.91% | |||
| 2017-06-26 | textminer | 2,399.8179 | 24.80% | 1,577.1395 | 50.08% | |||
| 2017-06-30 | onceAgain | 2,744.6813 | 13.68% | 2,038.8800 | 29.53% |