List your company
ResearchFormerly WizWand

Scene Text Recognition on IIIT5K (test)

99.6Word AccuracyCheng et al.
94.60895.90497.298.496Mar 12, 2016Sep 11, 2017Mar 14, 2019Sep 13, 2020Mar 15, 2022Sep 15, 2023Mar 17, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Cheng et al.Lexicon Size=50, Citation ID=[6]2019.0199.6
Cheng et al. 2018Lexicon=50, Training Level=word-level2018.1199.6
Shi et al. 2018Lexicon=50, Training Level=word-level2018.1199.6
Bai et al. 2018Lexicon=50, Training Level=word-level and character-level2018.1199.5
CLIP4STR-LTraining data=Real2023.1299.43
SARLexicon=50, Training Level=word-level2018.1199.4
CLIP4STR-LTraining Data=3.3M real samples, Backbone=ViT-L2023.0599.4
Cheng et al. 2017Lexicon=50, Training Level=word-level and character-level2018.1199.3
MPSTRTraining Data=3.3M real samples2023.0599.2
CLIP4STR-BTraining Data=3.3M real samples, Backbone=ViT-B2023.0599.2
CLIP4STR-L*Training data=REBU-Syn2023.1299.13
CLIP4STR-B*Training data=REBU-Syn2023.1298.96
Cheng et al.Lexicon Size=50, Citation ID=[5]2019.0198.9
PARSeqTraining Data=3.3M real samples2023.0598.9
Shi et al. 2018Lexicon=1k, Training Level=word-level2018.1198.8
CLIP4STR-BTraining data=Real2023.1298.73
Yin et al.Lexicon Size=502019.0198.7
ABINetTraining data=Real2023.1298.6
ABINetTraining Data=3.3M real samples, Backbone=ResNet-452023.0598.6
MAERec-BTraining data=Union14M-L2023.1298.5
SARLexicon=1k, Training Level=word-level2018.1198.2
Cheng et al.Lexicon Size=1k, Citation ID=[6]2019.0198.1
Cheng et al. 2018Lexicon=1k, Training Level=word-level2018.1198.1
Wang and Hu 2017Lexicon=50, Training Level=word-level2018.1198
MaskOCR(ViT-L)Training data=MJ+ST2023.1298
MORANLexicon Size=502019.0197.9
Bai et al. 2018Lexicon=1k, Training Level=word-level and character-level2018.1197.9
ViTSTR-STraining data=Real2023.1297.9
PARSeqATraining data=Real, Reproduced=true2023.1297.87
Shi, Bai, and YaoLexicon Size=502019.0197.8
Yang et al.Lexicon Size=502019.0197.8
Shi, Bai, and Yao 2017Lexicon=50, Training Level=word-level2018.1197.8
Yang et al. 2017Lexicon=50, Training Level=word-level and character-level2018.1197.8
Liu et al.Lexicon Size=502019.0197.7
Liu et al. 2016Lexicon=50, Training Level=word-level2018.1197.7
Shi et al.Lexicon Size=502016.0397.6
DiG-ViT-BTraining data=Real2023.1297.6
Cheng et al. 2017Lexicon=1k, Training Level=word-level and character-level2018.1197.5
Jaderberg et al. [17]Lexicon Size=502016.0397.1
Jaderberg et al.Lexicon Size=50, Citation ID=[22]2019.0197.1
Jaderberg et al. 2015aLexicon=50, Training Level=word-level2018.1197.1
Liu et al. 2018Lexicon=50, Training Level=word-level and character-level2018.1197
PARSeqTraining Data=90K+ST, Params=23.8M, Time=11.8ms2023.0597
PARSeqATraining data=MJ+ST2023.1297
Lee and OsinderoLexicon Size=502019.0196.8
Cheng et al.Lexicon Size=1k, Citation ID=[5]2019.0196.8
Lee and Osindero 2016Lexicon=50, Training Level=word-level2018.1196.8
CDistNet w/o TPSTraining Data=90K+ST2021.1196.7
DiG-ViT-BTraining data=MJ+ST2023.1296.7
MATRNTraining data=MJ+ST2023.1296.6
RARE (SRN only)Lexicon Size=502016.0396.5
CDistNetTraining Data=90K+ST2021.1196.4
NRTR+TPS++Training Data=90K+ST, Params=35.5M, Time=218ms2023.0596.3
ABINet-LV+TPS++Training Data=90K+ST, Params=37.2M, Time=41.5ms2023.0596.3
ABINet + InfoBatchPruned %=26.6%2025.0396.3
RARELexicon Size=502016.0396.2
Shi et al.Lexicon Size=50, Citation ID=[42]2019.0196.2
MORANLexicon Size=1k2019.0196.2
Shi et al. 2016Lexicon=50, Training Level=word-level2018.1196.2
ABINet-LVTraining Data=90K+ST, Params=36.7M, Time=37.2ms2023.0596.2
ABINetYear=2021, Training Data=90K+ST2021.1196.2
ABINetTraining data=MJ+ST2023.1296.2
ABINet + SeTaPruned %=28.1%2025.0396.2
Yang et al.Lexicon Size=1k2019.0196.1
Yin et al.Lexicon Size=1k2019.0196.1
Yang et al. 2017Lexicon=1k, Training Level=word-level and character-level2018.1196.1
ABINetPruned %=0%2025.0396.1
ABINet + InfoBatchPruned %=38.1%2025.0395.9
ABINet + SeTaPruned %=40.4%2025.0395.9
VisionLANTraining Data=90K+ST, Params=32.8M, Time=28.0ms2023.0595.8
GTRTraining Data=90K+ST, Params=42.1M, Time=18.8ms2023.0595.8
VisionLANYear=2021, Training Data=90K+ST2021.1195.8
S-GTRYear=2022, Training Data=90K+ST2021.1195.8
VisionLANTraining data=MJ+ST2023.1295.8
PETRTraining data=MJ+ST2023.1295.8
MaskOCR(ViT-B)Training data=MJ+ST2023.1295.8
ABINet + InfoBatchPruned %=50.3%2025.0395.8
ABINet + SeTaPruned %=71.0%2025.0395.8
TextScannerTraining data=MJ+ST2023.1295.7
Wang and Hu 2017Lexicon=1k, Training Level=word-level2018.1195.6
Pren2DTraining Data=90K+ST+Real, Time=67.4ms2023.0595.6
Pren2DYear=2021, Training Data=90K+ST+Real2021.1195.6
Jaderberg et al. [16]Lexicon Size=502016.0395.5
Jaderberg et al.Lexicon Size=50, Citation ID=[21]2019.0195.5
RobustScannerLexicon-free=true, Character-level annotations=false2021.0695.4
SGBANetTraining Data=90K+ST2023.0595.4
Mask TextSpotterLexicon-free=true, Character-level annotations=true2021.0695.3
RobustScannerTraining Data=90K+ST2023.0595.3
RobustScannerYear=2020, Training Data=90K+ST2021.1195.3
SPINTraining Data=90K+ST2023.0595.2
JVSRYear=2021, Training Data=90K+ST2021.1195.2
Shi, Bai, and YaoLexicon Size=1k2019.0195
Shi, Bai, and Yao 2017Lexicon=1k, Training Level=word-level2018.1195
SARLexicon=None, Training Level=word-level2018.1195
MASTERLexicon-free=true, Character-level annotations=false2021.0695
SARLexicon-free=true, Character-level annotations=false2021.0695
RCEEDLexicon-free=true, Character-level annotations=false2021.0694.9
SRNBackbone=ResNet, Training Data=Synth90K+SynthText, Annotations=word-level, Lexicon=None2020.0394.8
SRNTraining Data=90K+ST, Params=49.3M, Time=26.9ms2023.0594.8
SRNYear=2020, Training Data=90K+ST2021.1194.8