mirror/tesseract
1
0
Fork 0
mirror of https://github.com/tesseract-ocr/tesseract.git synced 2025-08-01 10:27:01 +08:00
Code Issues Actions 10 Packages Projects Releases Wiki Activity

Update based on https://github.com/tesseract-ocr/tessdata_best/pull/17#issuecomment-374455979

Shreeshrii 2018-03-21 09:38:19 +05:30
parent 08f4e27d99
commit c4de8b2cc5
1 changed files with 7 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
Data-Files-in-tessdata_fast.md

@ -2,17 +2,20 @@
We have three sets of .traineddata files for tesseract on GitHub in three separate repositories. We have three sets of .traineddata files for tesseract on GitHub in three separate repositories.
* https://github.com/tesseract-ocr/tessdata_fast (September 15, 2017, use for OCR) * [tessdata_fast](https://github.com/tesseract-ocr/tessdata_fast) (Sep 2017) best "value for money" in speed vs accuracy, Integer models
* https://github.com/tesseract-ocr/tessdata_best (September 15, 2017, use for finetune training) * [tessdata_best](https://github.com/tesseract-ocr/tessdata_best) (Sep 2017) best results on the eval data, slower, Float models, can be used as base for finetune training
* https://github.com/tesseract-ocr/tessdata (November 2016, supports legacy tesseract engine also) * [tessdata](https://github.com/tesseract-ocr/tessdata) (Nov 2016) supports legacy tesseract engine also
When using the models in the **`tessdata_best`** and **`tessdata_fast`** repositories, only the new LSTM-based OCR engine is supported. The legacy tesseract engine is NOT supported with these files, so Tesseract's oem modes '0' and '2' won't work with them. When using the models in the **`tessdata_best`** and **`tessdata_fast`** repositories, only the new LSTM-based OCR engine is supported. The legacy tesseract engine is NOT supported with these files, so Tesseract's oem modes '0' and '2' won't work with them.
## Information specific to tessdata_fast ## Information specific to tessdata_fast
First, fast is trained with a spec that produces a smaller net than best. As a result of smaller model, the prediction will be faster.
Then, the float->int conversion is done, which further reduces the size of the model and makes it even faster if your CPU supports AVX2.
### Usage ### Usage
Most users will use **`tessdata_fast`** for OCR as that is what will be shipped as part of Linux distributions and will provide accurate and fast recognition. Most users will use **`tessdata_fast`** for OCR as that is what will be shipped as part of Debian and Ubuntu distributions and will provide accurate and fast recognition.
### Version string : 4.00.00alpha : [Network specification] ### Version string : 4.00.00alpha : [Network specification]