Optical Character Recognition on Japanese Horizontal
0.6Character Error RateEffOCR-C (Base)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| EffOCR-C (Base)Seq2Seq?=X, Transformer?=X, Pretraining=from scratch, Parameters=112.5 M2023.04 | 0.6 | 0.79 | |
| EffOCR-T (Base)Seq2Seq?=X, Transformer?=✓, Pretraining=from scratch, Parameters=101.8 M2023.04 | 0.9 | 0.19 | |
| EffOCR-C (Small)Seq2Seq?=X, Transformer?=X, Pretraining=from scratch, Parameters=9.3 M2023.04 | 1 | 19.46 | |
| PaddleOCR SVTRSeq2Seq?=X, Transformer?=X, Pretraining=fine-tuned, Parameters=11 M2023.04 | 3.2 | — | |
| Baidu OCRSeq2Seq?=?, Transformer?=?, Pretraining=off-the-shelf, Parameters=?2023.04 | 6 | — | |
| EasyOCR CRNNPretraining=fine-tuned, Parameters=3.8 M2023.04 | 8.2 | — | |
| PaddleOCR SVTRSeq2Seq?=X, Transformer?=X, Pretraining=off-the-shelf, Parameters=11 M2023.04 | 8.5 | 13.34 | |
| PaddleOCR SVTRSeq2Seq?=X, Transformer?=X, Pretraining=from scratch, Parameters=11 M2023.04 | 9.7 | — | |
| EasyOCR CRNNPretraining=from scratch, Parameters=3.8 M2023.04 | 13.2 | — | |
| Google Cloud Vision OCRSeq2Seq?=?, Transformer?=?, Pretraining=off-the-shelf, Parameters=?2023.04 | 17.3 | — | |
| EasyOCR CRNNPretraining=off-the-shelf, Parameters=3.8 M2023.04 | 19.1 | 33.55 | |
| Tesseract OCR (Best)Seq2Seq?=X, Pretraining=off-the-shelf, Parameters=1.4 M2023.04 | 102.1 | 4.9 |