method: AntFin-Cascade Mask R-CNN2023-02-23
Authors: Yangkun Lin, Tao Xu
Affiliation: Ant Group
Description: Our detector is based on Cascade Mask R-CNN. We use ConvNeXt-B as backbone. SynthText800k and
VISD10k are used to pretrain, and then we finetune on ArT, ICDAR2019-MLT and part of LSVT with multi-scale training. Multi-scale testing is used to get the result.
method: I3CL2021-07-05
Authors: Jian Ye, Jing Zhang, Juhua Liu, Bo Du and Dacheng Tao
Affiliation: Wuhan University SigmaLab, JD Explore Academy
Email: leaf-yej@whu.edu.cn
Description: A arbitrary-shaped scene text detector based on Mask R-CNN. In this result, we use ResNeSt-101 as the backbone. Multi-scale training and testing are applied to get the final result. Our training datasets contain SynthText (pretrain), ArT, ICDAR2019-MLT, and part of LSVT.
method: MEGVII_Detection2019-04-30
Authors: Feng Wang, Li Hu, Peize Sun, Enze Xie, Wenhai Wang
Description: We proposed a method that use mask-rcnn to predict the segmentation as our boxes. Besides, we use senet-152 as our baseline and we also use multi-scale training and testing to improve the result.
Date | Method | Recall | Precision | Hmean | |||
---|---|---|---|---|---|---|---|
2023-02-23 | AntFin-Cascade Mask R-CNN | 76.02% | 79.41% | 77.68% | |||
2021-07-05 | I3CL | 73.68% | 79.35% | 76.41% | |||
2019-04-30 | MEGVII_Detection | 70.56% | 82.49% | 76.06% | |||
2019-12-17 | Tencent TEG OCR | 73.78% | 77.85% | 75.76% | |||
2019-11-04 | Sogou_OCR | 71.65% | 80.27% | 75.72% | |||
2020-05-21 | DuXiaoman_OCR | 71.42% | 79.04% | 75.04% | |||
2020-04-22 | Mask R-CNN | 70.75% | 77.84% | 74.12% | |||
2023-08-09 | SRFormer (ResNet50-#1seg) | 67.07% | 78.53% | 72.35% | |||
2024-01-18 | LRANet | 68.07% | 76.79% | 72.17% | |||
2019-04-29 | ArtDet-v2 | 66.63% | 78.32% | 72.01% | |||
2022-04-19 | TextBPN++(ResNet-50 with DCN) | 68.35% | 74.94% | 71.49% | |||
2023-07-08 | CPNText-DETR(resnet-50) | 68.17% | 74.90% | 71.38% | |||
2019-04-30 | CUTeOCR | 65.13% | 78.79% | 71.31% | |||
2020-10-01 | TextFuseNet (ResNeXt-101) | 65.94% | 77.40% | 71.21% | |||
2022-04-21 | I3CL(ViTAEv2-S) | 67.86% | 74.52% | 71.03% | |||
2022-07-11 | DPText-DETR (ResNet-50) | 66.85% | 75.26% | 70.81% | |||
2019-05-01 | NJU-ImagineLab | 65.04% | 76.55% | 70.33% | |||
2023-03-08 | TD-PPIoU | 68.41% | 71.99% | 70.16% | |||
2023-11-29 | ESRNet | 65.48% | 74.80% | 69.83% | |||
2019-04-26 | baseline_polygon | 65.41% | 71.59% | 68.36% | |||
2022-10-31 | TD-PPIoU (Long-Pretrain) | 63.93% | 73.28% | 68.29% | |||
2022-03-25 | TextBPN++(ResNet-50) | 62.78% | 71.68% | 66.94% | |||
2021-03-26 | TextFuseNet (ResNet-50) | 60.98% | 72.55% | 66.27% | |||
2021-04-08 | AutoCV | 62.68% | 69.58% | 65.95% | |||
2019-04-30 | DMText_art | 58.60% | 75.38% | 65.94% | |||
2019-04-29 | SRCB_Art | 61.15% | 69.95% | 65.25% | |||
2019-04-30 | A scene text detection method based on maskrcnn | 57.84% | 74.82% | 65.24% | |||
2019-04-29 | Sg_ptd | 59.15% | 72.23% | 65.04% | |||
2019-04-30 | Fudan-Supremind Detection v3 | 61.63% | 68.21% | 64.76% | |||
2019-04-28 | CLTDR | 58.14% | 72.83% | 64.66% | |||
2019-04-28 | Alibaba-PAI | 62.00% | 67.01% | 64.41% | |||
2023-05-15 | dp_pq_nn | 57.53% | 71.71% | 63.84% | |||
2023-05-15 | dp_nn | 58.23% | 68.29% | 62.86% | |||
2019-04-24 | fdu_ai | 53.48% | 71.27% | 61.11% | |||
2019-04-30 | CCISTD | 53.40% | 71.37% | 61.09% | |||
2021-04-28 | NN_Chinese_and_euro6 | 54.78% | 68.16% | 60.74% | |||
2019-05-01 | TextMask_V1 | 62.09% | 59.23% | 60.63% | |||
2019-04-30 | MaskRCNN_Text | 56.10% | 65.92% | 60.61% | |||
2019-04-30 | TEXT_SNIPER | 57.52% | 61.32% | 59.36% | |||
2019-04-30 | MaskDet | 55.43% | 63.22% | 59.07% | |||
2019-04-30 | Mask RCNN | 62.71% | 55.82% | 59.07% | |||
2019-04-16 | Art_test_baseline_task1 | 53.25% | 61.04% | 56.88% | |||
2019-04-30 | TMIS | 45.80% | 73.81% | 56.53% | |||
2021-04-23 | HOCRA | 54.30% | 58.86% | 56.49% | |||
2019-04-29 | CRAFT | 53.14% | 59.55% | 56.16% | |||
2019-04-22 | MFTD: Mask Filters for Text Detection | 52.41% | 59.92% | 55.92% | |||
2019-04-30 | QAQ | 48.86% | 64.49% | 55.60% | |||
2019-04-25 | Art detect by vivo | 47.44% | 67.00% | 55.55% | |||
2019-04-29 | PAT-S.Y | 48.46% | 61.53% | 54.22% | |||
2021-04-28 | NN_euro6 | 43.21% | 71.39% | 53.84% | |||
2019-04-30 | DMCA | 50.33% | 54.32% | 52.25% | |||
2019-04-26 | TP | 42.22% | 63.94% | 50.86% | |||
2019-04-22 | mask rcnn | 43.67% | 58.76% | 50.11% | |||
2019-04-28 | Improved Progressive scale expansion Net | 41.79% | 60.70% | 49.50% | |||
2019-05-01 | Unicamp-SRBR-PN-1 | 42.90% | 50.67% | 46.46% | |||
2019-04-27 | TextCohesion_1 | 34.80% | 54.26% | 42.40% | |||
2021-04-29 | HOCRA_base | 29.36% | 72.45% | 41.78% | |||
2019-04-23 | 1 | 42.27% | 41.08% | 41.66% | |||
2019-04-26 | RAST: Robust Arbitrary Shape Text Detector | 27.36% | 54.86% | 36.51% | |||
2019-04-30 | EM-DATA | 27.96% | 38.01% | 32.22% | |||
2021-03-31 | inception baseline | 16.97% | 33.66% | 22.57% | |||
2019-04-30 | MSR | 0.06% | 0.08% | 0.07% |