Text Detection on ICDAR 2015 (test)
91.55F1 ScoreCharNet H-88 MS
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| CharNet H-88 MSInference Scale=Multi-Scale, Params=89.21 M, Backbone=Hourglass-882019.10 | 91.55 | 92.65 | 90.47 | — | |
| ESTextSpotterParadigm=Supervised, Data Setting=Fully Labeled2025.04 | 91 | — | — | — | |
| CharNet H-88Inference Scale=Single Scale, Params=89.21 M, Backbone=Hourglass-882019.10 | 90.97 | 91.98 | 89.99 | — | |
| CharNet R-50 MSInference Scale=Multi-Scale, Params=26.48 M, Backbone=ResNet-502019.10 | 90.16 | 89.44 | 90.9 | — | |
| DeepSoloBackbone=ViTAEv2-S, External Data=Synth150K+MLT17+IC13+Total-Text+TextOCR2022.11 | 90.1 | 92.4 | 87.9 | — | |
| CharNet H-57 MSInference Scale=Multi-Scale, Params=34.96 M, Backbone=Hourglass-572019.10 | 90.06 | 88.74 | 91.43 | — | |
| TESTRRepresentation=Polygon2022.04 | 90 | 90.3 | 89.7 | — | |
| TESTRLexicon source=Specific lexicon [2], Character-level annotations=false2022.07 | 90 | 90.3 | 89.7 | — | |
| TESTRExternal Data=Synth150K+MLT17+Total-Text2022.11 | 90 | 90.3 | 89.7 | — | |
| DeepSoloBackbone=ResNet-50, External Data=Synth150K+MLT17+IC13+Total-Text2022.11 | 90 | 92.8 | 87.4 | — | |
| TESTRParadigm=Supervised, Data Setting=Fully Labeled2025.04 | 90 | — | — | — | |
| DeepSoloParadigm=Supervised, Data Setting=Fully Labeled2025.04 | 90 | — | — | — | |
| Supervised baselineParadigm=Semi-supervised, Data Setting=Fully Labeled2025.04 | 90 | — | — | — | |
| SemiETSParadigm=Semi-supervised, Data Setting=Fully Labeled2025.04 | 90 | — | — | — | |
| FOTS R-50 MSInference Scale=Multi-Scale, Params=34.98 M2019.10 | 89.84 | 91.85 | 87.92 | — | |
| DeepSoloBackbone=ResNet-50, External Data=Synth150K+MLT17+IC13+Total-Text+TextOCR2022.11 | 89.8 | 92.5 | 87.2 | — | |
| CharNet R-50Inference Scale=Single Scale, Params=26.48 M, Backbone=ResNet-502019.10 | 89.7 | 91.15 | 88.3 | — | |
| CharNetBackbone=ResNet-502022.04 | 89.7 | 91.2 | 88.3 | — | |
| CharNetLexicon source=Official, Character-level annotations=true2022.07 | 89.7 | 91.15 | 88.3 | — | |
| CharNetExternal Data=Synth800K2022.11 | 89.7 | 91.2 | 88.3 | — | |
| CharNet H-57Inference Scale=Single Scale, Params=34.96 M, Backbone=Hourglass-572019.10 | 89.66 | 90.45 | 88.88 | — | |
| Mean-Teacher*Paradigm=Semi-supervised, Data Setting=Fully Labeled2025.04 | 89.4 | — | — | — | |
| SRSTS v2Lexicon source=Official, Character-level annotations=false2022.07 | 89.35 | 93.43 | 85.6 | — | |
| TPSNetexternal dataset pretrain=true, end-to-end spotting training=true2021.10 | 89.1 | 90.5 | 87.8 | 11.6 | |
| Boundary TextSpotter2022.04 | 88.6 | 89.8 | 87.5 | — | |
| Boundaryexternal dataset pretrain=true, end-to-end spotting training=true2021.10 | 88.6 | 89.8 | 87.5 | — | |
| TPSNetexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 88.6 | 90.7 | 86.6 | 11.6 | |
| BoundaryLexicon source=Official, Character-level annotations=false2022.07 | 88.6 | 89.8 | 87.5 | — | |
| BoundaryExternal Data=Synth800K+IC13+Total-Text2022.11 | 88.6 | 89.8 | 87.5 | — | |
| SRSTSLexicon source=Official, Character-level annotations=false2022.07 | 88.44 | 96.05 | 81.96 | — | |
| SRSTSExternal Data=Synth800K+Synth150K+COCO-Text+MLT172022.11 | 88.4 | 96.1 | 82 | — | |
| PGNet2022.04 | 88.2 | 91.8 | 84.8 | — | |
| MOSTexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 88.2 | 87.3 | 89.1 | 10 | |
| PGNetLexicon source=Official, Character-level annotations=false2022.07 | 88.2 | 91.8 | 84.8 | — | |
| PGNetExternal Data=Synth800K+Total-Text2022.11 | 88.2 | 91.8 | 84.8 | — | |
| ABCNet v22022.04 | 88.1 | 90.4 | 86 | — | |
| ABCNetV2external dataset pretrain=true, end-to-end spotting training=true2021.10 | 88.1 | 90.4 | 86 | — | |
| ABCNet v2Lexicon source=Specific lexicon [2], Character-level annotations=true2022.07 | 88.1 | 90.4 | 86 | — | |
| ABCNet v2External Data=Synth150K+MLT172022.11 | 88.1 | 90.4 | 86 | — | |
| ABCNet v2Paradigm=Supervised, Data Setting=Fully Labeled2025.04 | 88.1 | — | — | — | |
| FOTS2022.04 | 88 | 91 | 85.2 | — | |
| FOTS2018.06 | 87.99 | 91 | 85.17 | 7.5 | |
| FOTS R-50Inference Scale=Single Scale, Params=34.98 M2019.10 | 87.99 | 91 | 85.17 | — | |
| FOTSLexicon source=Official, Character-level annotations=false2022.07 | 87.99 | 91 | 85.17 | — | |
| TextDragonExternal Data=Synth800K2022.11 | 87.9 | 92.5 | 83.8 | — | |
| Soft-Teacher*Paradigm=Semi-supervised, Data Setting=Fully Labeled2025.04 | 87.9 | — | — | — | |
| TextDragonLexicon source=Official, Character-level annotations=false2022.07 | 87.88 | 92.45 | 83.75 | — | |
| TPSNetexternal dataset pretrain=false, end-to-end spotting training=false2021.10 | 87.7 | 90.5 | 85.1 | 11.6 | |
| ATRRexternal dataset pretrain=false, end-to-end spotting training=false2021.10 | 87.6 | 89.2 | 86 | 10 | |
| Semi-DETR*Paradigm=Semi-supervised, Data Setting=Fully Labeled2025.04 | 87.6 | — | — | — | |
| Qin et al.Lexicon source=Official, Character-level annotations=false2022.07 | 87.52 | 89.36 | 85.75 | — | |
| Unconstrained2022.04 | 87.5 | 89.4 | 85.8 | — | |
| PAN++Lexicon source=Official, Character-level annotations=false2022.07 | 87.5 | 91.4 | 83.9 | — | |
| TextNet2022.04 | 87.4 | 89.4 | 85.4 | — | |
| TextNetLexicon source=Official, Character-level annotations=false2022.07 | 87.37 | 89.42 | 85.41 | — | |
| DBexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 87.3 | 91.8 | 83.2 | 12 | |
| PSENetOutput map scale=1/22018.06 | 87.21 | 89.3 | 85.22 | 7.88 | |
| LOMOexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 87.2 | 91.3 | 83.5 | 3.4 | |
| Text Perceptron2022.04 | 87.1 | 92.3 | 82.5 | — | |
| Mask-TTDexternal dataset pretrain=false, end-to-end spotting training=false2021.10 | 87.1 | 86.6 | 87.6 | — | |
| TextPerceptronLexicon source=Official, Character-level annotations=false2022.07 | 87.1 | 92.3 | 82.5 | — | |
| CRAFTSLexicon source=Official, Character-level annotations=true2022.07 | 87.1 | 89 | 85.3 | — | |
| TextPerceptronExternal Data=Synth800K2022.11 | 87.1 | 92.3 | 82.5 | — | |
| CRAFTSExternal Data=Synth800K+IC132022.11 | 87.1 | 89 | 85.3 | — | |
| PSENetOutput map scale=1/12018.06 | 87.08 | 88.71 | 85.51 | 2.33 | |
| He et al. MSInference Scale=Multi-Scale2019.10 | 87 | 87 | 86 | — | |
| He et al.2022.04 | 87 | 87 | 86 | — | |
| MaskTextSpotterexternal dataset pretrain=true, end-to-end spotting training=true2021.10 | 87 | 86.6 | 87.3 | — | |
| Mask Textspotter v2Lexicon source=Specific lexicon [2], Character-level annotations=false2022.07 | 87 | 86.6 | 87.3 | — | |
| CRAFTexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 86.9 | 89.8 | 84.3 | 8.6 | |
| SASTexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 86.9 | 86.7 | 87.1 | — | |
| ContourNetexternal dataset pretrain=false, end-to-end spotting training=false2021.10 | 86.9 | 87.6 | 86.1 | 3.5 | |
| SAEexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 86.6 | 88.3 | 85 | — | |
| DRRGNexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 86.6 | 88.5 | 84.7 | 3.5 | |
| FCENetexternal dataset pretrain=false, end-to-end spotting training=false2021.10 | 86.2 | 90.1 | 82.6 | — | |
| Mask TextSpotterInference Scale=Single Scale2019.10 | 86 | 91.6 | 81 | — | |
| ProposedExtraData=false2018.11 | 85.9 | 90.8 | 81.5 | — | |
| PSENetOutput map scale=1/42018.06 | 85.88 | 87.98 | 83.87 | 12.38 | |
| PSENetexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 85.7 | 86.9 | 84.5 | 1.6 | |
| GLASSExternal Data=Synth800K2022.11 | 85.7 | 86.9 | 84.5 | — | |
| GLASSParadigm=Supervised, Data Setting=Fully Labeled2025.04 | 85.7 | — | — | — | |
| Unbiased Teacher v2*Paradigm=Semi-supervised, Data Setting=Fully Labeled2025.04 | 85.5 | — | — | — | |
| IncepTextExtraData=false2018.11 | 85.3 | 90.5 | 80.6 | — | |
| TTS (poly)Paradigm=Supervised, Data Setting=Fully Labeled2025.04 | 85.2 | — | — | — | |
| SLPR2018.06 | 84.5 | 85.5 | 83.6 | — | |
| Lyu et al.2018.06 | 84.3 | 89.5 | 79.7 | 3.6 | |
| FTSN2018.06 | 84.14 | 88.65 | 80.07 | — | |
| FTSNExtraData=true2018.11 | 84.1 | 88.6 | 80 | — | |
| SwinTextSpotterParadigm=Supervised, Data Setting=Fully Labeled2025.04 | 83.9 | — | — | — | |
| PANexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 82.9 | 84 | 81.9 | 26.1 | |
| TextSnakeexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 82.6 | 84.9 | 80.4 | 1.1 | |
| R2CNN2018.06 | 82.54 | 85.62 | 79.68 | — | |
| R2CNNInference Scale=Single Scale2019.10 | 82.54 | 85.62 | 79.68 | — | |
| R2CNNExtraData=true2018.11 | 82.5 | 85.6 | 79.7 | — | |
| MSRexternal dataset pretrain=true, end-to-end spotting training=false2021.10 | 82.3 | 86.6 | 78.4 | — | |
| DDRExtraData=true2018.11 | 81 | 82 | 80 | — | |
| EAST2018.06 | 80.72 | 83.27 | 78.33 | 6.52 | |
| EASTInference Scale=Single Scale2019.10 | 80.72 | 83.27 | 78.33 | — | |
| EASTExtraData=false2018.11 | 80.7 | 83.2 | 78.3 | — | |
| WordSup2018.06 | 78.16 | 79.33 | 77.03 | — |