method: SituTech_OCR2021-03-11

Authors: Kui Lyu, Chuanhe Liu

Affiliation: Beijing Situ Vision Technologies Co. Ltd

Email: lvkui@situdata.com

Description: In this work, we design an elegant text detection model. Our detector is similar to DBNet, but there are some difference. More specifically, we have introduced an advanced detector backbone, a classic network EfficientDet, with flexible scales and stronger ability to extract features. Another breakthrough is that we optimized the label generation strategy in DBNet. In the original work, the positive area generation and the expansion of the positive area to the bounding box used the Vatti clipping algorithm, which is less robust with different area perimeter ratios. We optimized this function to make the label transform between positive area and bounding box more reasonable.

If you have any questions, please contact us.
SituAIgorithm Team, Beijing Situ Vision Technologies Co. Ltd

method: TH2020-04-19

Authors: Tsinghua University and Hyundai Motor Group AIRS Company

Email: Shanyu Xiao: xiaosy19@mails.tsinghua.edu.cn

Description: We have built an end-to-end scene text spotter based on Mask R-CNN & Transformer. The ResNeXt-101 backbone and multiscale training/testing are used.

method: Sogou_OCR2019-11-11

Authors: Xudong Rao, Lulu Xu, Long Ma, Xuefeng Su

Description: An arbitrary-shaped text detection method based on Mask R-CNN, we use resnext-152 as our backbone, multi-scale training and testing are adopted to get the final results.

Ranking Table

Description Paper Source Code
DateMethodHmeanPrecisionRecallAverage Precision
2021-03-11SituTech_OCR84.80%93.41%77.64%72.43%
2020-04-19TH84.67%89.54%80.30%77.90%
2019-11-11Sogou_OCR83.72%89.60%78.57%76.23%
2019-06-04Tencent-DPPR Team83.61%87.52%80.05%77.33%
2019-06-04Tencent-DPPR Team (Method_v0.3)83.61%87.57%79.99%77.28%
2019-06-04multi-stage_text_detector_v483.59%87.75%79.80%70.00%
2019-06-03Tencent-DPPR Team (Method_v0.2)83.55%87.78%79.72%77.05%
2019-06-04multi-stage_text_detector_v283.32%87.04%79.91%69.51%
2019-06-04multi-stage_text_detector_v383.30%87.15%79.78%69.49%
2019-06-03multi-stage_text_detector83.25%87.47%79.42%69.44%
2019-06-03NJU-ImagineLab(v3)83.07%87.85%78.79%76.21%
2019-05-30PMTD82.53%87.47%78.12%75.80%
2019-05-27Tencent-DPPR Team (Method_v0.1)81.88%89.41%75.52%72.87%
2019-05-29maskrcnn++ result80.35%82.64%78.19%64.64%
2019-05-29IC_RL80.11%82.97%77.44%64.29%
2019-06-02A two-stage text detector based on cascade rcnn(using total 10000 images of mlt19)78.38%82.26%74.85%71.27%
2019-05-31A two-stage text detector based on cascade rcnn78.11%82.89%73.85%70.31%
2021-02-04NCU_MSP78.00%83.08%73.51%60.82%
2019-06-03mm-maskrcnn_v276.79%84.73%70.21%67.44%
2019-06-04TH-DL-v276.64%84.55%70.09%64.44%
2019-06-03TH-DL-v176.59%84.51%70.03%64.35%
2019-05-27TH-DL76.53%84.70%69.80%64.07%
2020-05-30NCU75.80%78.54%73.25%57.06%
2020-10-16Drew75.42%83.71%68.63%65.03%
2019-05-26two stage text detector75.04%82.61%68.74%65.29%
2019-06-03sot74.24%79.96%69.28%65.94%
2019-06-02DISTILLED CRAFT72.94%81.22%66.19%59.16%
2019-06-03text-mountain71.95%72.12%71.77%51.90%
2019-06-03CRAFTS70.86%81.42%62.73%56.63%
2019-06-04Unicamp-SRBR-MLT2019-PELEETEXT70.81%81.58%62.54%59.01%
2019-06-03RRPN69.56%77.71%62.95%58.07%
2019-06-04Unicamp-SRBR-MLT2019-FUSION-PSENET-PELEETEXT68.56%77.00%61.79%56.03%
2019-05-28CRAFTS(Initial)68.11%79.51%59.56%54.50%
2019-06-04Lomin OCR67.65%71.62%64.09%57.95%
2019-06-03 NXB OCR65.96%70.59%61.90%43.72%
2019-05-24PSENet_v165.83%73.52%59.59%52.73%
2019-05-27MLT2019 ETD64.36%78.71%54.44%42.93%
2019-05-27CLTDR63.53%77.20%53.97%41.63%
2020-10-23MEAST_V3_23_Oct61.64%70.02%55.04%39.81%
2020-10-07MEAST_V2_8_oct61.49%70.45%54.55%39.66%
2019-05-27NXB OCR61.31%71.84%53.48%38.48%
2020-09-18MEAST_V158.07%68.84%50.22%34.87%
2019-06-03TP58.01%77.59%46.32%37.26%
2019-05-28Unicamp-SRBR-MLT2019-S151.00%75.22%38.58%35.30%
2019-06-04Cyberspace47.09%69.48%35.61%26.17%
2019-05-28PydBox-TextDetector29.79%59.56%19.86%11.83%
2020-12-15DSIT-UOA23.43%28.42%19.93%7.83%
2019-05-05AAAA0.02%0.07%0.01%0.00%
2019-05-27Unicamp-SRBR-MLT2019-S10.00%0.03%0.00%0.00%
2019-06-01tsinghuaee51_MLT20190.00%0.04%0.00%0.00%
2019-05-274Paradigm-Data-Intelligence0.00%0.00%0.00%0.00%

Ranking Graphic

Ranking Graphic