Last active
June 10, 2023 23:51
-
-
Save bertsky/e17ae89b9757a4676a3e209a20d1d396 to your computer and use it in GitHub Desktop.
OCR-D workflow for coverage tests, esp. CUDA support
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| set -e | |
| # select first CUDA device (in case there are multiple, which may fail due to [a recent Tensorflow problem](https://github.com/qurator-spk/eynollah/issues/99)) | |
| export CUDA_VISIBLE_DEVICES=0 | |
| # check we are not running into [this bug](https://github.com/shapely/shapely/issues/1598) | |
| python3 -c "from shapely.geometry import Polygon; import torch; torch.randn(10).cuda()" | |
| # validate CUDA support is working in TF and Torch (not an exhaustive test) | |
| python3 -c "import torch; print(torch.cuda.is_available())" | |
| python3 -c "import tensorflow as tf; print(tf.test.is_gpu_available())" | |
| # get some test data | |
| git clone https://github.com/OCR-D/assets/ | |
| cd assets/data/SBB0000F29300010000/data/ | |
| # download models and run some processors (not for result quality, only coverage) | |
| ocrd resmgr download ocrd-sbb-binarize default-2021-03-09 | |
| ocrd-sbb-binarize -I OCR-D-IMG -O OCR-D-BIN -P model default-2021-03-09 | |
| ocrd resmgr download ocrd-detectron2-segment Jambo-sudo_X101.pth | |
| ocrd resmgr download ocrd-detectron2-segment Jambo-sudo_X101.yaml | |
| ocrd-detectron2-segment -p $(python -c "import ocrd_detectron2; print(ocrd_detectron2.__path__[0])")/presets_Jambo-sudo_X101.json -I OCR-D-BIN -O OCR-D-SEG | |
| ocrd-typegroups-classifier -I OCR-D-IMG -O FONT | |
| ocrd resmgr download ocrd-eynollah-segment default | |
| ocrd-eynollah-segment -P models default -I OCR-D-IMG -O OCR-D-SEG2 | |
| ocrd resmgr download ocrd-calamari-recognize qurator-gt4histocr-1.0 | |
| ocrd-calamari-recognize -I OCR-D-SEG2 -O OCR-D-OCR -P checkpoint_dir qurator-gt4histocr-1.0 -P textequiv_level glyph | |
| ocrd resmgr download ocrd-kraken-segment blla.mlmodel | |
| ocrd-kraken-segment -I OCR-D-BIN -O OCR-D-SEG3 -P device cuda:0 | |
| ocrd-cis-ocropy-resegment -I OCR-D-SEG3 -O OCR-D-SEG3X -P method baseline | |
| ocrd resmgr download ocrd-kraken-recognize reichsanzeiger.mlmodel | |
| ocrd-kraken-recognize -I OCR-D-SEG3X -O OCR-D-OCR2 -P model reichsanzeiger.mlmodel -P device cuda:0 | |
| wget "https://git.informatik.uni-leipzig.de/ocr-d/cor-asv-ann-models/-/raw/master/s2s.gt4histocr.s-%C5%BF.d2.w0512.adam.attention.stateless.variational-dropout.char.transfer-lm.h5" | |
| ocrd resmgr download -n s2s.gt4histocr.s-ſ.d2.w0512.adam.attention.stateless.variational-dropout.char.transfer-lm.h5 ocrd-cor-asv-ann-process s2s.gt4histocr.s-ſ.d2.w0512.adam.attention.stateless.variational-dropout.char.transfer-lm.h5 | |
| ocrd-cor-asv-ann-process -I OCR-D-OCR -O OCR-D-COR -P model_file s2s.gt4histocr.s-ſ.d2.w0512.adam.attention.stateless.variational-dropout.char.transfer-lm.h5 | |
| ocrd-anybaseocr-crop -I OCR-D-BIN -O OCR-D-CROP | |
| ocrd-skimage-denoise -I OCR-D-CROP -O OCR-D-DEN | |
| ocrd-cis-ocropy-segment -I OCR-D-DEN -O OCR-D-SEG4 -P level-of-operation page | |
| ocrd-segment-evaluate -I OCR-D-SEG,OCR-D-SEG2 -O OCR-D-SEGEVAL | |
| ocrd resmgr download ocrd-tesserocr-recognize frak2021.traineddata | |
| ocrd-tesserocr-recognize -I OCR-D-SEG2 -O OCR-D-OCR3 -P model frak2021 | |
| ocrd resmgr download ocrd-cis-ocropy-recognize LatinHist.pyrnn.gz | |
| ocrd-cis-ocropy-recognize -I OCR-D-SEG2 -O OCR-D-OCR4 -P model LatinHist.pyrnn.gz -P textequiv_level glyph | |
| ocrd-cor-asv-ann-align -I OCR-D-OCR,OCR-D-OCR3,OCR-D-OCR4 -O OCR-D-OCR5 | |
| ocrd-cor-asv-ann-evaluate -I OCR-D-OCR,OCR-D-OCR3,OCR-D-OCR4,OCR-D-OCR5 -O OCR-D-OCREVAL | |
| ocrd-page-transform -I OCR-D-OCR4 -O OCR-D-OCR4X -P xsl page-textequiv-lines-to-regions.xsl | |
| ocrd-fileformat-transform -I OCR-D-OCR4X -O TXT -P from-to "page text" -P script-args level=region pb="$(echo -e \v)" | |
| ocrd-fileformat-transform -I OCR-D-OCR4X -O FULLTEXT -P from-to "page alto" -P script-args "--no-check-border --dummy-word" |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment