Update README.md
Browse files
README.md
CHANGED
|
@@ -23,10 +23,6 @@ This benchmark contains paired Persian document images and UTF-8 transcription
|
|
| 23 |
targets for OCR evaluation. Each JSONL row references one image under
|
| 24 |
`bench_data/images/` and contains its transcription in the `text` field.
|
| 25 |
|
| 26 |
-
The dataset is organized in the same benchmark-oriented style as
|
| 27 |
-
[olmOCR-bench](https://huggingface.co/datasets/allenai/olmOCR-bench): a
|
| 28 |
-
`bench_data/` directory, a JSONL test manifest, and an evaluation configuration.
|
| 29 |
-
|
| 30 |
## Schema
|
| 31 |
|
| 32 |
- `image`: image path relative to the repository
|
|
@@ -36,9 +32,4 @@ The dataset is organized in the same benchmark-oriented style as
|
|
| 36 |
- `text`: reference transcription
|
| 37 |
- `language`: `fa`
|
| 38 |
- `checked`: local-label status
|
| 39 |
-
- `notes`: reviewer notes
|
| 40 |
-
|
| 41 |
-
## License and privacy
|
| 42 |
-
|
| 43 |
-
Review the source-image rights and remove or redact personal, financial, and
|
| 44 |
-
government-identifying information before making this repository public.
|
|
|
|
| 23 |
targets for OCR evaluation. Each JSONL row references one image under
|
| 24 |
`bench_data/images/` and contains its transcription in the `text` field.
|
| 25 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
## Schema
|
| 27 |
|
| 28 |
- `image`: image path relative to the repository
|
|
|
|
| 32 |
- `text`: reference transcription
|
| 33 |
- `language`: `fa`
|
| 34 |
- `checked`: local-label status
|
| 35 |
+
- `notes`: reviewer notes
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|