Post-correction of OCR results
The proposed task concerns post-correcting OCR results of Polish-language books, which were published in the years 1791-1998. [ver. 2.0.3]
poleval-2021 ocr ocr-correction
Git repo URL: https://github.com/poleval/2021-ocr-correction.git / Branch: main
Run git clone --single-branch https://github.com/poleval/2021-ocr-correction.git -b main to get the challenge data
Browse at https://github.com/poleval/2021-ocr-correction/tree/main
Leaderboard
| # | submitter | when | ver. | description | test-A WER | test-B WER | × | |
|---|---|---|---|---|---|---|---|---|
| 1 | None | 2021-09-01 12:58 | 2.0.1 poleval-2021 | XXLv1 | 3.725 | 3.744 | 5 | |
| 2 | Adam Mickiewicz University & Applica.ai | 2021-10-23 21:42 | 2.0.2 after-poleval-2021 | plt5/allegro large after 20 epochs of fine-tuning after-poleval-2021 language-model | 4.079 | 4.239 | 5 | |
| 3 | eNeLPol UJ AGH | 2021-09-04 18:56 | 2.0.1 poleval-2021 | ED 3 pl | N/A | 4.302 | 4 | |
| 4 | Adam Mickiewicz University & Applica.ai | 2021-09-06 06:39 | 2.0.1 poleval-2021 | uml language-model test | 5.338 | 5.328 | 5 | |
| 5 | niezależny | 2021-09-06 11:59 | 2.0.1 poleval-2021 | v.1 | N/A | 7.208 | 1 | |
| 6 | Samurai Labs, Wrocław University of Science and Technology | 2021-09-02 18:43 | 2.0.1 poleval-2021 | Heuristics no-ml | 8.063 | 8.217 | 10 |