method: TencentAILab.v22018-03-03
Authors: Jingchao Zhou, Tianlin Gao, Zhifeng Li
Description: Adopt a Faster-RCNN architecture with rotation-rpn to detect both axis aligned and inclined text lines, then post-process the detect candidates carefully with tranditional OCR technology.
method: TencentAILab.v12018-01-19
Authors: Wei Zhang, Xinyu Li, Wu Liu, Pengrui Duan, Zhineng Chen, Xiaochun Cao, Zhifeng Li
Description: This run is based on Faster-RCNN, with rotational region proposals. By IIE of CAS, BUPT and TencentAILab.
method: Tencent-OCR+(v2)2017-06-20
Authors: Yi Li, Weichen Zhang, Hui Song, Chunchao Guo, Ming Liu, Hongfa Wang, Lei Xiao
Description: Data Platform Department, Tencent. Our method is based on Faster-RCNN architecture. In the Faster-RCNN, we adapt a rotation-rpn which is helpful to detect inclined text region.
Description Paper Source Code
| Date | Method | Average Precision | |||
|---|---|---|---|---|---|
| 2018-03-03 | TencentAILab.v2 | 93.15% | |||
| 2018-01-19 | TencentAILab.v1 | 91.45% | |||
| 2017-06-20 | Tencent-OCR+(v2) | 89.05% | |||
| 2017-06-20 | Tencent-OCR+(v1) | 82.74% | |||
| 2017-06-28 | MMD | 78.80% | |||
| 2017-06-30 | SCUT_DLVClab | 77.28% | |||
| 2019-06-07 | PSENet | 70.18% | |||
| 2019-06-24 | CRAFT | 68.78% | |||
| 2017-06-30 | onceAgain | 56.41% | |||
| 2019-06-14 | FOTS | 48.78% | |||
| 2019-05-03 | EAST | 47.43% | |||
| 2017-11-03 | dd | 45.76% | |||
| 2017-06-29 | textminer | 15.88% |