Handwritten Text Recognition on IAM A partition (test)
2.75CERDiaz et al., 2021
Evaluation Results
| Method | Links | |
|---|---|---|
| Diaz et al., 2021Training Data=Internal + IAM, Architecture=S-Attn/CTC, External LM=Yes, Decoding Strategy=greedy decoding2022.05 | 2.75 | |
| TrOCR(LARGE) 2021Training Data=Synthetic + IAM, Architecture=Transformer, External LM=No, Decoding Strategy=greedy decoding, #param(M)=5582022.05 | 2.89 | |
| Diaz et al., 2021Training Data=Internal + IAM, Architecture=Transformer w/ CNN, External LM=No, Decoding Strategy=greedy decoding2022.05 | 2.96 | |
| Bluche and Messina, 2017Training Data=Synthetic + IAM, Architecture=GCRNN/ CTC, External LM=Yes, Decoding Strategy=greedy decoding2022.05 | 3.2 | |
| Ingle et al., 2019Training Data=Internal + IAM, Architecture=GRCL, External LM=No, Decoding Strategy=greedy decoding, #param(M)=10.62022.05 | 4 | |
| Kang et al., 2020Training Data=Synthetic + IAM, Architecture=Transformer w/ CNN, External LM=No, Decoding Strategy=greedy decoding, #param(M)=1002022.05 | 4.67 | |
| Michael et al., 2019Training Data=Internal + IAM, Architecture=LSTM/LSTM w/Attn, External LM=No, Decoding Strategy=greedy decoding2022.05 | 4.87 | |
| Easter2.0Training Data=IAM (6k), Architecture=1DCNN-SE/CTC, External LM=No, Decoding Strategy=greedy decoding, #param(M)=6.12022.05 | 6.21 | |
| Wang et al.Training Data=IAM (6k), Architecture=DAN, FCN/GRU, External LM=No, Decoding Strategy=greedy decoding2022.05 | 6.4 | |
| Kang et al., 2020Training Data=IAM (6k), Architecture=Transformer w/ CNN, External LM=No, Decoding Strategy=greedy decoding, #param(M)=1002022.05 | 7.62 | |
| Chaudhary et al., 2021Training Data=IAM (6k), Architecture=FCN/CTC, External LM=No, Decoding Strategy=greedy decoding, #param(M)=282022.05 | 9.8 | |
| Ingle et al., 2019Training Data=IAM (6k), Architecture=GRCL, External LM=No, Decoding Strategy=greedy decoding, #param(M)=10.62022.05 | 14.1 |