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

Authors: Longhuang Wu, Shangxuan Tian, Chang Liu, Wenjie Cai, Jiachen Li, Sicong Liu, Haoxi Li, Chunchao Guo, Hongfa Wang, Hongkai Chen, Qinglin lu, Chun Yang, Xucheng Yin, Lei Xiao

Description: We are Tencent-DPPR (Data Platform Precision Recommendation) team. Our method follows the framework of Mask R-CNN that employs mask to detect multi-oriented scene texts. We use the MLT-19 and the MSRA-TD500 dataset to train our text detector, and we also apply a multi-scale training approach during training. To obtain the final ensemble results, we combined two different backbones and different multi-scale testing approaches.

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.

Ranking Table

Description Paper Source Code
DateMethodHmeanPrecisionRecallAverage Precision
2021-03-11SituTech_OCR63.33%46.38%99.81%46.85%
2019-05-27Tencent-DPPR Team (Method_v0.1)58.16%42.03%94.37%81.25%
2020-04-19TH56.51%40.13%95.45%77.12%
2023-05-22DeepSolo++ (ResNet-50)55.99%39.70%94.93%91.58%
2019-11-11Sogou_OCR54.68%38.17%96.35%91.79%
2019-06-03TH-DL-v153.60%37.58%93.44%74.95%
2019-05-27TH-DL53.56%38.03%90.48%69.44%
2019-06-03CRAFTS53.50%38.00%90.32%60.64%
2019-06-04TH-DL-v253.44%37.40%93.54%74.64%
2019-06-03Tencent-DPPR Team (Method_v0.2)52.53%36.18%95.81%82.61%
2019-06-04multi-stage_text_detector_v452.42%36.32%94.18%33.51%
2019-06-04Tencent-DPPR Team (Method_v0.3)52.19%35.85%95.88%82.39%
2019-06-04Tencent-DPPR Team52.14%35.81%95.81%82.36%
2019-06-03multi-stage_text_detector51.53%35.65%92.92%32.31%
2019-06-04multi-stage_text_detector_v351.22%35.25%93.66%32.26%
2019-06-02DISTILLED CRAFT51.21%36.55%85.49%28.08%
2019-06-04multi-stage_text_detector_v251.20%35.14%94.25%32.25%
2019-06-03TP51.07%36.09%87.31%34.14%
2022-11-02ESTextSpotter50.83%34.64%95.41%92.18%
2019-05-30PMTD50.53%34.49%94.46%90.94%
2019-06-03NJU-ImagineLab(v3)50.45%34.25%95.74%90.42%
2019-06-03mm-maskrcnn_v250.20%34.27%93.82%90.35%
2019-05-28Unicamp-SRBR-MLT2019-S149.41%36.02%78.65%63.17%
2019-05-31A two-stage text detector based on cascade rcnn47.54%31.76%94.51%91.13%
2019-05-24PSENet_v147.19%32.66%85.01%50.03%
2019-05-26two stage text detector46.64%30.99%94.20%89.16%
2019-06-02A two-stage text detector based on cascade rcnn(using total 10000 images of mlt19)46.16%30.53%94.58%90.87%
2021-02-04NCU_MSP45.79%30.52%91.67%28.08%
2019-05-28CRAFTS(Initial)44.87%30.16%87.59%57.06%
2019-05-27CLTDR44.21%29.17%91.24%25.80%
2020-10-16Drew43.88%29.71%83.93%70.96%
2019-06-03sot43.24%27.96%95.36%92.29%
2019-06-04Unicamp-SRBR-MLT2019-PELEETEXT42.20%27.37%92.09%82.80%
2019-05-29IC_RL41.63%26.80%93.28%24.04%
2019-05-27MLT2019 ETD41.60%27.03%90.25%23.66%
2019-05-29maskrcnn++ result40.50%25.81%94.01%23.34%
2019-06-03RRPN40.00%25.46%93.23%85.25%
2019-06-04Unicamp-SRBR-MLT2019-FUSION-PSENET-PELEETEXT39.72%25.28%92.54%81.61%
2020-05-30NCU36.45%22.81%90.72%20.22%
2019-05-27NXB OCR35.93%22.88%83.74%24.40%
2019-06-04Lomin OCR35.22%21.80%91.67%73.18%
2020-10-07MEAST_V2_8_oct35.09%21.88%88.52%28.42%
2019-06-03text-mountain34.37%21.25%89.91%18.43%
2020-10-23MEAST_V3_23_Oct33.24%20.42%89.39%27.36%
2023-05-30TD-PPIoU29.48%17.57%91.50%79.06%
2019-06-03 NXB OCR25.83%15.39%80.26%12.07%
2019-06-04Cyberspace23.17%14.77%53.72%9.20%
2019-05-28PydBox-TextDetector21.87%17.88%28.15%5.04%
2020-12-15DSIT-UOA7.04%3.83%43.23%10.30%
2019-05-05AAAA0.02%0.01%0.05%0.00%
2019-05-274Paradigm-Data-Intelligence0.00%0.00%0.00%0.00%
2019-05-27Unicamp-SRBR-MLT2019-S10.00%0.00%0.00%0.00%
2019-06-01tsinghuaee51_MLT20190.00%0.00%0.00%0.00%

Ranking Graphic

Ranking Graphic