OCR Correction on English OCR Dataset
3CERByT5
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ByT5Correction Stage=After (Post-OCR), Synthetic Data Method=Method 4 (Glyph Similarity)2024.08 | 3 | 7.74 | |
| OCR BaselineCorrection Stage=Before (Raw OCR)2024.08 | 4.96 | 15.43 |
Loading the SOTA2 catalog…