method: SogouMM2021-01-08
Authors: MiaoWang,DanDan Li, Tao Wei,Hongyuan Zhang,FengGu,XuDong Rao
Description: Our method is based on 2D-attention, we simply use ResNet as backbone and a tailored 2D-attention module is applied. The result is generated by a single model without ensemble tricks.
method: Tencent TEG OCR2020-03-15
Authors: Pei Xu, Shan Huang, Shen Huang, Qi Ju.
Description: We reimplemented the standalone recognition method according to the end-to-end text spotting code released by the Mask TextSpotter[TPAMI]. It is a seq-to-seq method based on 2D attention. We synthesize curved text images for pretraining by the method of VGG synthtext. We add public dataset including icdar2013-2015, CUTE, SVT, IIIT5k, RCTW2017, LSVT,ctw to finetune and don't use any private data.
method: SogouMM2019-11-08
Authors: Xu Liu, Hongyuan Zhang, Jin Chen, Tao Wei
Description: Our method is based on 2D-attention, we simply use ResNet as backbone and a tailored 2D-attention module is applied. The result is generated by a single model without ensemble tricks.
Date | Method | Result | Total words | Correct words | |||
---|---|---|---|---|---|---|---|
2021-01-08 | SogouMM | 80.38% | 35284 | 28362 | |||
2020-03-15 | Tencent TEG OCR | 80.36% | 35284 | 28355 | |||
2019-11-08 | SogouMM | 80.13% | 35284 | 28272 | |||
2019-11-06 | Sogou_OCR | 78.14% | 35284 | 27571 | |||
2019-10-24 | hw-noah-lab-gts-CV-team | 76.60% | 35284 | 27029 | |||
2019-04-30 | PKU Team Zero | 74.30% | 35284 | 26216 | |||
2019-04-30 | CUTeOCR | 73.91% | 35284 | 26078 | |||
2019-05-01 | CRAFT (Preprocessing) + TPS-ResNet | 73.87% | 35284 | 26063 | |||
2019-04-29 | serial_rec | 72.89% | 35284 | 25717 | |||
2019-04-29 | NPU-ASGO | 71.82% | 35284 | 25341 | |||
2019-04-30 | TPS-ResNet | 71.77% | 35284 | 25324 | |||
2020-10-15 | transformer_v1 | 71.36% | 35284 | 25179 | |||
2019-05-01 | CIGIT & XJTLU | 70.73% | 35284 | 24956 | |||
2019-05-01 | Attention based method for scene text recognition | 70.39% | 35284 | 24838 | |||
2019-05-01 | Attention based method for arbitrary-shaped scene text recognition | 70.28% | 35284 | 24796 | |||
2020-07-09 | Lenovo-MI-Lab OCR | 69.22% | 35284 | 24424 | |||
2019-04-27 | Ensemble and post processes | 69.15% | 35284 | 24398 | |||
2019-04-29 | CAN-icdar | 69.01% | 35284 | 24350 | |||
2019-04-30 | CSN-ED | 67.32% | 35284 | 23753 | |||
2019-05-01 | Alchera AI | 66.81% | 35284 | 23573 | |||
2019-04-30 | Irregular Text Recognizer with Attention Mechanism | 64.45% | 35284 | 22739 | |||
2019-04-28 | class_5435_rotate | 63.86% | 35284 | 22534 | |||
2019-04-26 | Irregular Text Recognition with Direction Classification and a Rectification Network | 61.91% | 35284 | 21846 | |||
2019-04-29 | MatchCRNN | 58.03% | 35284 | 20476 | |||
2019-04-21 | Arbitrary shape scene text recognition based on CNN and Attention Enhanced Bi-directional LSTM | 56.09% | 35284 | 19792 | |||
2019-05-01 | Fudan-Supremind Recognition | 50.56% | 35284 | 17838 | |||
2019-04-30 | LCT_OCR(中国科学院信息工程研究所) | 47.31% | 35284 | 16693 | |||
2019-05-01 | So Cold 2.0 | 45.30% | 35284 | 15982 | |||
2019-04-16 | Art_test_baseline_task2 | 43.74% | 35284 | 15433 | |||
2019-04-29 | task2x | 38.08% | 35284 | 13437 |