qurator-spk/eynollah
1
0
Fork 0
mirror of https://github.com/qurator-spk/eynollah.git synced 2026-07-11 22:29:29 +02:00
Code Issues Projects Releases Wiki Activity

Commit graph

  • ca8bbe8506
         Merge c680dae2d1 into 2e3f45c300 Robert Sachunsky 2026-07-10 15:24:01 +00:00
  • c680dae2d1 ModelZoo for ONNX backend: allow setting execution providers via env Robert Sachunsky 2026-07-10 17:22:50 +02:00
  • 39e054e718 train: remove tf-specifics and weird __getitem__ def from transformer-ocr setup integrating_trocr_and_torch_ensembling_and_updating_characters_list-refactor kba 2026-07-09 19:23:21 +02:00
  • affddd6c85 train: make preprocess_imgs_ocr work for transformer-ocr kba 2026-07-09 19:12:09 +02:00
  • b2f3a8f2d8 Merge branch 'fix-0.8-modelzoo-and-predictor-kba0709' into integrating_trocr_and_torch_ensembling_and_updating_characters_list-refactor kba 2026-07-09 17:32:23 +02:00
  • b1f2f43051 upgrade tf2onnx fork dep, remove spurious tf-data dep fix-0.8-modelzoo-and-predictor-kba0709 kba 2026-07-09 17:30:48 +02:00
  • 492fcbacb7 switch to fork of tf2onnx kba 2026-07-09 15:17:11 +02:00
  • b9ba43b444 training: require tf2onnx and pin ml_dtypes >= 0.5 kba 2026-07-09 14:37:51 +02:00
  • c9c14ed83d Docker: update to ONNX base image (in Makefile, too) Robert Sachunsky 2026-07-08 14:04:19 +02:00
  • 98e28ca9f4 Docker: fixup cuDNN installation after OCR extra Robert Sachunsky 2026-07-08 03:05:07 +02:00
  • 5354583913 update readme Robert Sachunsky 2026-07-08 02:56:11 +02:00
  • 8cc8c28471 ModelZoo for ONNX backend: configure TRT cache path from XDG env Robert Sachunsky 2026-07-08 02:55:25 +02:00
  • 32568a590f Docker: update to ONNX base image Robert Sachunsky 2026-07-08 02:54:34 +02:00
  • 171a8a3161 move TF+Keras dependencies to training extra, replace by ONNX+TRT Robert Sachunsky 2026-07-07 22:56:36 +02:00
  • 1567df1379 remove numba dependency (previously used to free CUDA memory) Robert Sachunsky 2026-07-07 22:09:46 +02:00
  • 16943f70b4 models (ViT backbone) iterate extract_patches over batch dim… Robert Sachunsky 2026-07-02 20:58:22 +02:00
  • 1b27c7390f training.convert/ONNX: run strict shape inference and check model Robert Sachunsky 2026-07-02 20:56:55 +02:00
  • 5e531ab006 get_textlines_of_textregion_sorted: fix be61875 Robert Sachunsky 2026-07-01 18:27:46 +02:00
  • 01e69a0e22 predictor for OCR models: work around ONNX bug … Robert Sachunsky 2026-06-26 02:42:51 +02:00
  • 948d841a7d training.models for cnn-rnn-ocr: make ONNX convertible… Robert Sachunsky 2026-06-26 02:21:47 +02:00
  • 45168178dc ModelZoo ONNX backend: log configured top provider (backend) Robert Sachunsky 2026-06-26 02:13:39 +02:00
  • ef47f0ef09 training convert: only add characters_org.txt if it exists Robert Sachunsky 2026-06-26 02:11:59 +02:00
  • 42a3751e63 ModelZoo ONNX: avoid verbose logging Robert Sachunsky 2026-06-19 22:01:39 +02:00
  • 0bfbbfdc80 training.metrics: allow module init without TFA Robert Sachunsky 2026-06-19 22:01:02 +02:00
  • eb4cae9dee training.models for cnn-rnn-ocr: avoid Conv1D(..channels_first..) Robert Sachunsky 2026-06-16 17:28:10 +02:00
  • dfa651ef8a predictor: show full stacktrace before passing the exception over Robert Sachunsky 2026-06-12 22:21:06 +02:00
  • 89ce9de6fa musicregion is added to pagexml to label vahidrezanezhad 2026-02-24 15:46:15 +01:00
  • 303bdfe0e7 Amiri font which works for both arabic and latin vahidrezanezhad 2026-02-24 13:55:45 +01:00
  • d2123a2746 FIXME: get label for decoration without type attribute vahidrezanezhad 2026-02-24 01:39:12 +01:00
  • 499e3d0715 trocr inference is integrated - works on CPU cause seg fault on GPU vahidrezanezhad 2026-02-18 15:04:54 +01:00
  • a11c833fc1 bug fix: layout visualization vahidrezanezhad 2026-02-16 11:50:39 +01:00
  • d0b3bb419f extracting ocr textline images and text: vertical lines threshold has changed to 1.4 vahidrezanezhad 2026-02-10 14:32:23 +01:00
  • 4776ea9fc4 torch model ensembling is integrated vahidrezanezhad 2026-02-04 21:16:08 +01:00
  • aba0138216 generate or update list of characters in the case of cnn-rnn ocr training vahidrezanezhad 2026-02-03 20:20:20 +01:00
  • 7f86a55ccb integrating transformer ocr vahidrezanezhad 2026-02-03 19:45:50 +01:00
  • e9839a8b54 makefile to reload models: avoid ONNX conversion for cnn-rnn-ocr too Robert Sachunsky 2026-06-12 15:00:36 +02:00
  • 19504cb932 makefile to reload models: add target for SavedModel Keras format Robert Sachunsky 2026-06-12 14:59:38 +02:00
  • 60c9f4786c ModelZoo device selection: warn if model category still unmatched… Robert Sachunsky 2026-06-12 14:58:32 +02:00
  • 94082bc64a ModelZoo TF-Serving backend: deal with buggy .inputs signature… Robert Sachunsky 2026-06-12 14:56:44 +02:00
  • 45c92eada2 models w/ multiple inputs yield a tuple for .input_shape Robert Sachunsky 2026-06-12 14:55:46 +02:00
  • 08946067ac ModelZoo ONNX backend: handle multiple inputs, too Robert Sachunsky 2026-06-12 14:54:51 +02:00
  • 9d2412080f training.models for cnn-rnn-ocr: fix config names for height/width… Robert Sachunsky 2026-06-12 14:52:23 +02:00
  • 4181e03bc9 training convert --rebuild for cnn-rnn-ocr: override charset file… Robert Sachunsky 2026-06-12 14:48:47 +02:00
  • 33734d3eeb
         Merge 28a559c710 into 2e3f45c300 Konstantin Baierer 2026-06-11 19:07:25 +02:00
  • 28a559c710 Merge remote-tracking branch 'bertsky/fix-0.8-modelzoo-and-predictor' into integrating_trocr_and_torch_ensembling_and_updating_characters_list kba 2026-06-11 19:07:19 +02:00
  • 6df11d92d8 Merge branch 'integrating_trocr_and_torch_ensembling_and_updating_characters_list' kba 2026-06-11 18:59:33 +02:00
  • bed7fe526b remove (half-implemented) page_alto functionality kba 2026-06-11 17:04:56 +02:00
  • 348ac95ad3 Eynollah_ocr: drop fixed input sizes… Robert Sachunsky 2026-06-03 20:59:00 +02:00
  • 24c7d4c277 update trocr smoke test, add cnnrnn ocr smoke test Robert Sachunsky 2026-06-03 20:58:05 +02:00
  • 27ca9733db ModelZoo ONNX backend for inference: support multi-input or -output Robert Sachunsky 2026-06-03 20:57:02 +02:00
  • 38fe4d33ad Predictor for multi-input models: present as list instead of tuple… Robert Sachunsky 2026-06-03 20:56:00 +02:00
  • 4e7e1c06b9 trocr viarant for Predictor runtime: no model size for input_shape… Robert Sachunsky 2026-06-03 20:51:56 +02:00
  • f447a9f248 trocr: move preprocessor and decoder into model object, too… Robert Sachunsky 2026-06-03 03:41:44 +02:00
  • d2f2a1e06b Eynollah_ocr: correctly handle min_conf, improve writer… Robert Sachunsky 2026-06-03 00:43:46 +02:00
  • 8ffc4ed8d3 Eynollah_ocr: adapt to inference model, improve and simplify… Robert Sachunsky 2026-06-02 21:20:06 +02:00
  • a391ee24e6 Predictor: handle multi-input and/or multi-output cases Robert Sachunsky 2026-06-02 21:18:22 +02:00
  • c79b73dcc8 cnn-rnn-ocr: move CTC decoder and string decoder to inference model… Robert Sachunsky 2026-06-02 20:26:42 +02:00
  • 13f2f81c45 ModelZoo: support inference with ONNX/TensorRT… Robert Sachunsky 2026-05-28 18:08:08 +02:00
  • f833a516e7 training: add CLI command convert… Robert Sachunsky 2026-05-28 17:48:21 +02:00
  • 62b55a3809 train params: drop reload_weights, re-use dir_of_start_model… Robert Sachunsky 2026-05-28 17:42:55 +02:00
  • 093030f503 train/models: move all model builders to models.get_model()… Robert Sachunsky 2026-05-28 17:37:45 +02:00
  • faef1967f8 models.cnn_rnn_ocr_model: add inference option, drop model name Robert Sachunsky 2026-05-28 17:32:02 +02:00
  • 2e3f45c300
         Merge pull request #220 from dokempf/remove-vim-swap-file main Clemens Neudecker 2026-05-28 11:41:21 +02:00
  • e17712eacd Remove accidentally committed vim swap file Dominic Kempf 2026-05-28 11:27:41 +02:00
  • c4a7eec5b3 models: cosmetics Robert Sachunsky 2026-05-27 01:58:21 +02:00
  • 9801129aa6 estimate_skew_contours: ensure retval is always float Robert Sachunsky 2026-05-22 12:37:07 +02:00
  • 26afc5ddab ModelZoo: ensure exported TensorShape is converted to plain tuple Robert Sachunsky 2026-05-22 12:35:44 +02:00
  • 0836230c6b utils_ocr: avoid module-level import of TF Robert Sachunsky 2026-05-21 22:50:53 +02:00
  • f3a93983c0 ModelZoo: add ocr key for memory_limit Robert Sachunsky 2026-05-21 22:50:13 +02:00
  • ea41dcae1d trocr: use beam search instead of greedy decoding Robert Sachunsky 2026-05-21 17:52:27 +02:00
  • 074753a98e ModelZoo: fix Torch device selection Robert Sachunsky 2026-05-21 17:25:53 +02:00
  • 000e4ac8d8 trocr: extract confidence, too Robert Sachunsky 2026-05-21 17:25:39 +02:00
  • f3649adbf2 trocr: apply do_not_mask_with_textline_contour here, too Robert Sachunsky 2026-05-21 17:23:11 +02:00
  • 1d67e65f11 trocr: simplify, batch over entire page… Robert Sachunsky 2026-05-21 15:48:21 +02:00
  • d50bd7c650 trocr: avoid warnings by passing clean_up_tokenization_spaces=False Robert Sachunsky 2026-05-21 14:20:51 +02:00
  • f9f9130dbb do_order_of_regions: remove redundant+overcautious assertion Robert Sachunsky 2026-05-21 03:21:36 +02:00
  • bf7ec0233d ModelZoo.load_model: use memory_limit instead of memory_growth… Robert Sachunsky 2026-05-21 02:43:34 +02:00
  • 94a5e9da14 ModelZoo.load_model: avoid attempting to load exported models as Keras models (which causes a warning), but switch to TF-Serving import right away Robert Sachunsky 2026-05-21 02:41:19 +02:00
  • 7f2bf715df ModelZoo.load_model: fix loading exported vs saved models Robert Sachunsky 2026-05-21 02:39:59 +02:00
  • 3de1407d18 drop unnecessary TF / Torch imports Robert Sachunsky 2026-05-21 02:38:20 +02:00
  • bdfebd2c70 reload_weights: save() → export() w/ serve() inference Robert Sachunsky 2026-05-19 03:20:24 +02:00
  • 86adaf299a training.models.transformer_block: tf.reshape → Keras Reshape layer Robert Sachunsky 2026-05-19 03:17:31 +02:00
  • 9efce5e9f2 Predictor.shutdown: use join() instead of terminate() Robert Sachunsky 2026-05-19 03:16:15 +02:00
  • ffe5cdc519 ModelZoo.shutdown: drop extra del (already done by shutdown()) Robert Sachunsky 2026-05-19 02:09:49 +02:00
  • 481c286da9 ModelZoo.load_model: no XLA compilation Robert Sachunsky 2026-05-19 02:08:14 +02:00
  • f329e10a80 test_layout: rm ignored --allow_scaling option Robert Sachunsky 2026-05-12 20:04:41 +02:00
  • 17b311441a model_zoo: also parse comma/colon syntax for device in Torch case Robert Sachunsky 2026-05-12 20:02:40 +02:00
  • be4fe8c263 contour: drop unused functions depending on rotation_image_new() Robert Sachunsky 2026-05-12 19:04:37 +02:00
  • 87cce6c963 CLI tests: add opt-in envvar EYNOLLAH_OPTIONS for device selection, model directory etc. Robert Sachunsky 2026-05-12 19:03:32 +02:00
  • 1ed633bc25 test_model_zoo: adapt (load_models instead of load_model) Robert Sachunsky 2026-05-12 19:02:43 +02:00
  • 21ecb043f7 CLIs: move --device option to group level Robert Sachunsky 2026-05-12 18:41:21 +02:00
  • 7ed1a1ebac CLIs: allow -h and show defaults uniformly, harmonise help, drop remaining redundant negative options Robert Sachunsky 2026-05-12 18:34:56 +02:00
  • cd62f13872 eynollah_ocr: make work again, re-use Eynollah base class… Robert Sachunsky 2026-05-12 18:31:18 +02:00
  • ded668a256 model_zoo: fix clash between Predictor and direct (OCR) use-cases… Robert Sachunsky 2026-05-12 18:17:43 +02:00
  • 98e6fbbcbb mbreorder: make work again, re-use Eynollah base class Robert Sachunsky 2026-05-11 11:30:39 +02:00
  • 7e8b9311d3 Revert "test_model_zoo: fix calls" Robert Sachunsky 2026-05-19 03:32:37 +02:00
  • a1449da1d1 Revert "fix model loading in mb_ro and ocr" Robert Sachunsky 2026-05-19 03:32:19 +02:00
  • 1df32eba87 CD: base docker image: typo {,v}3.13.0 kba 2026-05-11 13:41:30 +02:00
  • d7337a3080 CD: base docker image on versioned ocrd/core-cuda-tf2:v3.13.0 kba 2026-05-11 13:38:36 +02:00
  • e612db2bb1 📦 v0.8.0 v0.8.0 kba 2026-05-11 13:16:30 +02:00