Text Detection on Total-Text
93.9PrecisionDeepSolo
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DeepSoloBackbone=ResNet-50, External Data=Synth150K, FPS (report)=17, FPS (A100)=172022.11 | 93.9 | 82.1 | 87.6 | — | — | |
| DeepSoloBackbone=Res-50, External Data=Synth150K, FPS (report)=17, FPS (A100)=172023.05 | 93.9 | 82.1 | — | — | 87.6 | |
| TESTRScale=L: 16002022.07 | 93.4 | 81.4 | 86.9 | — | — | |
| TESTRExternal Data=Synth150K+MLT17, FPS (report)=5.3, FPS (A100)=12.12022.11 | 93.4 | 81.4 | 86.9 | — | — | |
| TESTRExternal Data=Synth150K+MLT17, FPS (report)=5.3, FPS (A100)=12.12023.05 | 93.4 | 81.4 | — | — | 86.9 | |
| TESTRFPS=5.32025.02 | 93.4 | 81.4 | 86.9 | — | — | |
| SRSTS v2 (ours)Scale=S: 6402022.07 | 93.3 | 86.74 | 89.9 | — | — | |
| DeepSoloBackbone=ResNet-50, External Data=Synth150K+MLT17+IC13+IC15+TextOCR, FPS (report)=17, FPS (A100)=172022.11 | 93.2 | 84.6 | 88.7 | — | — | |
| DeepSoloBackbone=Res-50, External Data=Synth150K+MLT17+IC13+IC15+TextOCR, FPS (report)=17, FPS (A100)=172023.05 | 93.2 | 84.6 | — | — | 88.7 | |
| DeepSoloFPS=17.02025.02 | 93.2 | 84.6 | 88.7 | — | — | |
| DeepSoloBackbone=ResNet-50, External Data=Synth150K+MLT17+IC13+IC15, FPS (report)=17, FPS (A100)=172022.11 | 93.1 | 82.1 | 87.3 | — | — | |
| DeepSoloBackbone=Res-50, External Data=Synth150K+MLT17+IC13+IC15, FPS (report)=17, FPS (A100)=172023.05 | 93.1 | 82.1 | — | — | 87.3 | |
| DeepSoloBackbone=ViTAEv2-S, External Data=Synth150K+MLT17+IC13+IC15+TextOCR, FPS (report)=10, FPS (A100)=102022.11 | 92.9 | 87.4 | 90 | — | — | |
| DeepSoloBackbone=ViTAEv2-S, External Data=Synth150K+MLT17+IC13+IC15+TextOCR, FPS (report)=10, FPS (A100)=102023.05 | 92.9 | 87.4 | — | — | 90 | |
| DeepSolocode_source=open-source, FPS=17.02025.02 | 92.8 | 82.4 | 87.4 | — | — | |
| SRSTS v2-F (ours)Scale=S: 6402022.07 | 92.13 | 83.97 | 87.86 | — | — | |
| SRSTSExternal Data=Synth800K+Synth150K+COCO-Text+MLT17+ArT19+IC15, FPS (report)=18.72022.11 | 92 | 83 | 87.2 | — | — | |
| BridgeSpotterFPS=6.72025.02 | 92 | 86.5 | 89.2 | — | — | |
| SRSTS (ours)Scale=S: 6402022.07 | 91.99 | 82.96 | 87.24 | — | — | |
| PSENet (VLPT-STD)Pre-training=VLPT-STD2022.04 | 90.8 | 82 | 86.1 | — | — | |
| GLASSExternal Data=Synth800K, FPS (report)=32022.11 | 90.8 | 85.5 | 88.1 | — | — | |
| GLASSExternal Data=Synth800K, FPS (report)=32023.05 | 90.8 | 85.5 | — | — | 88.1 | |
| GLASSFPS=2.72025.02 | 90.8 | 85.5 | 88.1 | — | — | |
| ABPNetBackbone=ResNet50, External Data (Ext.)=true, Method Category=Non-real-time2021.11 | 90.7 | 85.2 | 87.9 | 10.7 | — | |
| OmniParser V2FPS=8.22025.02 | 90.6 | 88.2 | 89.4 | — | — | |
| ABCNet v22021.05 | 90.2 | 84.1 | — | — | 87 | |
| ABCNet v2Scale=S: 10002022.07 | 90.2 | 84.1 | 87 | — | — | |
| ABCNet v2External Data=Synth150K+MLT17, FPS (report)=10, FPS (A100)=14.92022.11 | 90.2 | 84.1 | 87 | — | — | |
| ABCNet v2External Data=Synth150K+MLT17, FPS (report)=10, FPS (A100)=14.92023.05 | 90.2 | 84.1 | — | — | 87 | |
| TPSNetExternal Data=Synth150K+MLT17, Backbone=Res-50-DCN, FPS (report)=9.32023.05 | 90.2 | 86.8 | — | — | 88.5 | |
| ABCNet v2FPS=10.02025.02 | 90.2 | 84.1 | 87 | — | — | |
| TPSNet2025.02 | 90.2 | 86.8 | 88.5 | — | — | |
| FAST-B-800Backbone=TextNet-B, External Data (Ext.)=true, Input Resolution (Shorter Side)=800, Method Category=Real-time2021.11 | 90 | 85.2 | 87.5 | 46 | — | |
| OmniParser V2 PTpre-training=PT, FPS=8.22025.02 | 90 | 87.6 | 88.8 | — | — | |
| CharNetBackbone=Hourglass-88, Multi-scale=false2019.10 | 89.9 | 81.7 | 85.6 | — | — | |
| PAN++Backbone=ResNet18, External Data (Ext.)=true, Method Category=Real-time2021.11 | 89.9 | 81 | 85.3 | 38.3 | — | |
| FAST-B-640Backbone=TextNet-B, External Data (Ext.)=true, Input Resolution (Shorter Side)=640, Method Category=Real-time2021.11 | 89.9 | 83.2 | 86.4 | 67.5 | — | |
| OmniParser V2 SPTpre-training=SPT, FPS=8.22025.02 | 89.7 | 88.5 | 89.1 | — | — | |
| FAST-B-512Backbone=TextNet-B, External Data (Ext.)=true, Input Resolution (Shorter Side)=512, Method Category=Real-time2021.11 | 89.6 | 82.4 | 85.8 | 93.2 | — | |
| CRAFTSScale=L: 1920, Character-level annotations=true2022.07 | 89.5 | 85.4 | 87.4 | — | — | |
| CRAFTSExternal Data=Synth800K+IC13, character_level_annotations=true2022.11 | 89.5 | 85.4 | 87.4 | — | — | |
| CRAFTSExternal Data=Synth800K+IC13, Character-level Annotations=true2023.05 | 89.5 | 85.4 | — | — | 87.4 | |
| CRAFTSFPS=8.82025.02 | 89.5 | 85.4 | 87.4 | — | — | |
| PANexternal_data=true, input_scale_short_side=512, backbone=ResNet-18, pre_training=SynthText2019.08 | 89.4 | 79.7 | 84.3 | 57.1 | — | |
| PANexternal_data=true, input_scale_short_side=640, backbone=ResNet-18, pre_training=SynthText2019.08 | 89.3 | 81 | 85 | 39.6 | — | |
| PANextra training data=true2021.04 | 89.3 | 81 | 85 | — | — | |
| FCENetbackbone=ResNet50, DCN=true, extra training data=false2021.04 | 89.3 | 82.5 | 85.8 | — | — | |
| PAN2021.05 | 89.3 | 81 | — | — | 85 | |
| PAN2022.02 | 89.3 | 81 | 85 | 39.6 | — | |
| PANBackbone=ResNet18, External Data (Ext.)=true, Method Category=Real-time2021.11 | 89.3 | 81 | 85 | 39.6 | — | |
| PSENetPre-training=STKM2022.04 | 89.2 | 79.9 | 84.3 | — | — | |
| TextFuseNetBackbone=ResNet101, External Data (Ext.)=true, Method Category=Non-real-time2021.11 | 89 | 85.3 | 87.1 | 3.3 | — | |
| DBNet++Backbone=ResNet-50, Input Height=8002022.02 | 88.9 | 83.2 | 86 | 28 | — | |
| DB++-R50Backbone=ResNet50, External Data (Ext.)=true, Method Category=Real-time2021.11 | 88.9 | 83.2 | 86 | 28 | — | |
| BoundaryScale=L: 11002022.07 | 88.9 | 85 | 87 | — | — | |
| BoundaryExternal Data=Synth800K+IC13+IC152022.11 | 88.9 | 85 | 87 | — | — | |
| BoundaryExternal Data=Synth800K+IC13+IC152023.05 | 88.9 | 85 | — | — | 87 | |
| Boundary2025.02 | 88.9 | 85 | 87 | — | — | |
| TextPerceptronScale=L: 13502022.07 | 88.8 | 81.8 | 85.2 | — | — | |
| TextPerceptronExternal Data=Synth800K2022.11 | 88.8 | 81.8 | 85.2 | — | — | |
| TextPerceptronExternal Data=Synth800K2023.05 | 88.8 | 81.8 | — | — | 85.2 | |
| TextPerceptron2025.02 | 88.8 | 81.8 | 85.2 | — | — | |
| CharNetBackbone=Hourglass-57, Multi-scale=false2019.10 | 88.6 | 81 | 84.6 | — | — | |
| LOMOBackbone=ResNet50, External Data (Ext.)=true, Method Category=Non-real-time2021.11 | 88.6 | 75.7 | 81.6 | 4.4 | — | |
| CharNet H-57Character-level annotations=true2022.07 | 88.6 | 81 | 84.6 | — | — | |
| CharNetExternal Data=Synth800K, character_level_annotations=true2022.11 | 88.6 | 81 | 84.6 | — | — | |
| CharNetExternal Data=Synth800K, Character-level Annotations=true2023.05 | 88.6 | 81 | — | — | 84.6 | |
| CharNetFPS=5.42025.02 | 88.6 | 81 | 84.6 | — | — | |
| PAN++Scale=S: 5122022.07 | 88.4 | 80.5 | 84.2 | — | — | |
| OmniParserFPS=2.62025.02 | 88.4 | 88.6 | 88.5 | — | — | |
| Mask TextSpotter2021.05 | 88.3 | 82.4 | — | — | 85.2 | |
| DBNetBackbone=ResNet-18, Input Height=8002022.02 | 88.3 | 77.9 | 82.8 | 50 | — | |
| DB-R18Backbone=ResNet18, External Data (Ext.)=true, Method Category=Real-time2021.11 | 88.3 | 77.9 | 82.8 | 50 | — | |
| FAST-S-512Backbone=TextNet-S, External Data (Ext.)=true, Input Resolution (Shorter Side)=512, Method Category=Real-time2021.11 | 88.3 | 81.7 | 84.9 | 115.5 | — | |
| CharNetBackbone=Hourglass-88, Multi-scale=true2019.10 | 88 | 85 | 86.5 | — | — | |
| PANexternal_data=false, input_scale_short_side=640, backbone=ResNet-182019.08 | 88 | 79.4 | 83.5 | 39.6 | — | |
| PANextra training data=false2021.04 | 88 | 79.4 | 83.5 | — | — | |
| ABCNetextra training data=true2021.04 | 87.9 | 81.3 | 84.5 | — | — | |
| CRAFTextra training data=true2021.04 | 87.6 | 79.9 | 83.6 | — | — | |
| LOMOextra training data=true2021.04 | 87.6 | 79.3 | 83.3 | — | — | |
| LOMO2021.05 | 87.6 | 79.3 | — | — | 83.3 | |
| CRAFT2021.05 | 87.6 | 79.9 | — | — | 83.6 | |
| LOMOMulti-scale testing=true2022.02 | 87.6 | 79.3 | 83.3 | — | — | |
| CRAFT2022.02 | 87.6 | 79.9 | 83.6 | — | — | |
| CRAFTBackbone=VGG16, External Data (Ext.)=true, Method Category=Non-real-time2021.11 | 87.6 | 79.9 | 83.6 | 4.8 | — | |
| SAME-NetBackbone=Swin-Trans.2026.05 | 87.44 | 88.48 | — | — | 87.96 | |
| FCENet†backbone=ResNet50, DCN=false, extra training data=false2021.04 | 87.4 | 79.8 | 83.4 | — | — | |
| DBNet++Backbone=ResNet-18, Input Height=8002022.02 | 87.4 | 79.6 | 83.3 | 48 | — | |
| FCENetBackbone=ResNet50, External Data (Ext.)=false, Method Category=Non-real-time2021.11 | 87.4 | 79.8 | 83.4 | — | — | |
| DB++-R18Backbone=ResNet18, External Data (Ext.)=true, Method Category=Real-time2021.11 | 87.4 | 79.6 | 83.3 | 48 | — | |
| CharNetBackbone=Hourglass-57, Multi-scale=true2019.10 | 87.3 | 85 | 86.1 | — | — | |
| DBextra training data=true2021.04 | 87.1 | 82.5 | 84.7 | — | — | |
| DB2021.05 | 87.1 | 82.5 | — | — | 84.7 | |
| DBNetBackbone=ResNet-50, Input Height=8002022.02 | 87.1 | 82.5 | 84.7 | 32 | — | |
| DB-R50Backbone=ResNet50, External Data (Ext.)=true, Method Category=Real-time2021.11 | 87.1 | 82.5 | 84.7 | 32 | — | |
| ContourNetextra training data=false2021.04 | 86.9 | 83.9 | 85.4 | — | — | |
| CounterNet2021.05 | 86.9 | 83.9 | — | — | 85.4 | |
| ContourNetBackbone=ResNet50, External Data (Ext.)=false, Method Category=Non-real-time2021.11 | 86.9 | 83.9 | 85.4 | 3.8 | — | |
| PGNetBackbone=FPN2026.05 | 86.8 | 85.5 | — | — | 86.1 | |
| PANexternal_data=false, input_scale_short_side=512, backbone=ResNet-182019.08 | 86.7 | 78.4 | 82.4 | 57.1 | — |