Handwriting Recognition on IAM
2.38CERDTrOCR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DTrOCRTraining Data=Synthetic + IAM, External LM=No2023.08 | 2.38 | — | |
| DTrOCRArchitecture=Decoder-Only Transform., Params=300M, Training Data=2B+ Synthetic Lines, Adaptation Cost=>500 GPU hrs2026.03 | 2.38 | — | |
| S-Attn/CTC + LMRect.=No, Encoder/Decoder=S-Attn/CTC + LM, Train Dataset=Internal + Public2021.04 | 2.75 | — | |
| Diaz et al.Architecture=S-Attn/CTC, Training Data=Internal + IAM, External LM=Yes2021.09 | 2.75 | — | |
| Diaz et al.Training Data=Internal + IAM, External LM=Yes2023.08 | 2.75 | — | |
| TrOCR_LARGEArchitecture=Transformer, Training Data=Synthetic + IAM, External LM=No2021.09 | 2.89 | — | |
| TrOCRLARGETraining Data=Synthetic + IAM, External LM=No2023.08 | 2.89 | — | |
| TrOCR-LargeArchitecture=End-to-End Transform., Params=558M, Training Data=684M Synthetic Lines, Adaptation Cost=300+ GPU hrs2026.03 | 2.89 | — | |
| TransformerRect.=No, Encoder/Decoder=Transformer, Train Dataset=Internal + Public2021.04 | 2.96 | — | |
| Diaz et al.Architecture=Transformer w/ CNN, Training Data=Internal + IAM, External LM=No2021.09 | 2.96 | — | |
| S-Attn/CTC + LMRect.=No, Encoder/Decoder=S-Attn/CTC + LM, Train Dataset=Internal2021.04 | 3.15 | — | |
| Bluche and MessinaRect.=No, Encoder/Decoder=GCRNN/CTC, Train Dataset=IAM (50k lexicon)2021.04 | 3.2 | — | |
| Bluche and MessinaArchitecture=GCRNN/CTC, Training Data=Synthetic + IAM, External LM=Yes2021.09 | 3.2 | — | |
| Bluche et al.Training Data=Synthetic + IAM, External LM=Yes2023.08 | 3.2 | — | |
| TrOCR_BASEArchitecture=Transformer, Training Data=Synthetic + IAM, External LM=No2021.09 | 3.42 | — | |
| TrOCR-BaseArchitecture=End-to-End Transform., Params=334M, Training Data=684M Synthetic Lines, Adaptation Cost=200+ GPU hrs2026.03 | 3.42 | — | |
| S-Attn/CTCRect.=No, Encoder/Decoder=S-Attn/CTC, Train Dataset=Internal + Public2021.04 | 3.53 | — | |
| Diaz et al.Architecture=S-Attn/CTC, Training Data=Internal + IAM, External LM=No2021.09 | 3.53 | — | |
| TransformerRect.=No, Encoder/Decoder=Transformer, Train Dataset=Internal2021.04 | 3.99 | — | |
| transformerLevel=line2025.04 | 3.99 | — | |
| TrOCR_SMALLArchitecture=Transformer, Training Data=Synthetic + IAM, External LM=No2021.09 | 4.22 | — | |
| DANIELLevel=page2025.04 | 4.38 | — | |
| MT-DANLevel=page2025.04 | 4.51 | — | |
| DANLevel=page2025.04 | 4.54 | — | |
| S-Attn/CTCRect.=No, Encoder/Decoder=S-Attn/CTC, Train Dataset=Internal2021.04 | 4.62 | — | |
| Kang et al.Rect.=No, Encoder/Decoder=Transformer, Train Dataset=IAM2021.04 | 4.67 | — | |
| Kang et al.Architecture=Transformer w/ CNN, Training Data=Synthetic + IAM, External LM=No2021.09 | 4.67 | — | |
| Kang et al.Training Data=Synthetic + IAM, External LM=No2023.08 | 4.67 | — | |
| Meta-DANLevel=page2025.04 | 4.79 | — | |
| Michael et al.Rect.=No, Encoder/Decoder=LSTM/LSTM w/Attn, Train Dataset=IAM2021.04 | 4.87 | — | |
| Michael et al.Architecture=LSTM/LSTM w/Attn, Training Data=IAM, External LM=No2021.09 | 4.87 | — | |
| Michael et al.Training Data=IAM, External LM=No2023.08 | 4.87 | — | |
| BART-BaseArchitecture=Linguistic Adapter, Params=140M, Training Data=1.5M Synthetic, Adaptation Cost=3 + 0.5 GPU hrs2026.03 | 5.18 | — | |
| T5-BaseArchitecture=Linguistic Adapter, Params=220M, Training Data=1.5M Synthetic, Adaptation Cost=3 + 0.5 GPU hrs2026.03 | 5.4 | — | |
| Detector BaselineArchitecture=Object Detector Only, Params=40M, Training Data=1M Synthetic, Adaptation Cost=3 GPU hrs2026.03 | 6.09 | — | |
| ByT5-BaseArchitecture=Linguistic Adapter, Params=582M, Training Data=1.5M Synthetic, Adaptation Cost=3 + 1.0 GPU hrs2026.03 | 6.35 | — | |
| Wang et al.Rect.=No, Encoder/Decoder=FCN/GRU, Train Dataset=MJ + ST2021.04 | 6.4 | — | |
| Wang et al.Architecture=FCN/GRU, Training Data=IAM, External LM=No2021.09 | 6.4 | — | |
| Wang et al.Training Data=IAM, External LM=No2023.08 | 6.4 | — | |
| ABINetTraining-free evaluation=true2023.11 | — | 61.57 | |
| ASTERTraining-free evaluation=true2023.11 | — | 52.5 | |
| E2STR-baseTraining-free evaluation=true, In-Context Learning=false2023.11 | — | 69.51 | |
| E2STR-ICLTraining-free evaluation=true, In-Context Learning=true2023.11 | — | 74.1 | |
| MAERecTraining-free evaluation=true2023.11 | — | 70.27 | |
| NRTRTraining-free evaluation=true2023.11 | — | 59.53 | |
| SARTraining-free evaluation=true2023.11 | — | 56.63 | |
| SATRNTraining-free evaluation=true2023.11 | — | 59.47 |