qurator-spk/eynollah
1
0
Fork 0
mirror of https://github.com/qurator-spk/eynollah.git synced 2025-10-26 23:34:13 +01:00
Code Issues Projects Releases Wiki Activity

Commit graph

  • 48266b1ee0 make training dependencies optional-dependencies of eynollah kba 2025-10-01 18:01:21 +02:00
  • 733af1e9a7 📝 update train/README.md, align with docs/train.md kba 2025-10-01 17:43:32 +02:00
  • 5725e4fd1f -Continue processing when num_col is None but textregions exist. -Convert marginal-only to main body if no main body is present. -Reset deskew angle to 0 when text region density (textregion area to page area) < 0.3 and angle > 45°. vahidrezanezhad 2025-10-01 15:58:03 +02:00
  • 4514d417a7 force GH markdown code block in list cneud 2025-10-01 01:16:25 +02:00
  • e027bc038e Update README.md cneud 2025-10-01 01:05:15 +02:00
  • 91d2a74ac9 remove redundant parentheses cneud 2025-10-01 00:38:01 +02:00
  • f2f93e0251 list literal is faster than using list constructor to create a new list cneud 2025-10-01 00:26:27 +02:00
  • 70af00182b mutable defaults are the source of all evil cneud 2025-10-01 00:20:18 +02:00
  • 1d0616eb69 comparisons to None should not use the equality operators cneud 2025-10-01 00:15:11 +02:00
  • 9ce127eb51 remove unnecessary backslash cneud 2025-10-01 00:04:53 +02:00
  • 558867eb24 fix typo cneud 2025-10-01 00:04:07 +02:00
  • 3aa7ad04fa 📝 update changelog Robert Sachunsky 2025-09-30 23:14:52 +02:00
  • f0de1adabf rm loky dependency Robert Sachunsky 2025-09-30 23:12:18 +02:00
  • 7daec392b9 Dockerfile: fix up CUDA installation for mixed TF/Torch Robert Sachunsky 2025-09-30 22:10:45 +02:00
  • ad129ed46c CI: remove OS from model cache keys Robert Sachunsky 2025-09-30 22:05:53 +02:00
  • c86e59f481 CI: update model key, split up cache restore/save Robert Sachunsky 2025-09-30 22:03:46 +02:00
  • a3d8197930 makefile: update model URL Robert Sachunsky 2025-09-30 21:50:21 +02:00
  • 61b20cc83d tests: switch from subtests to parametrize, use --isolate everywhere to free CUDA memory in between Robert Sachunsky 2025-09-30 19:20:35 +02:00
  • 375e0263d4 CNN-RNN OCR model: switch to 20250930 version (compatible with TF 2.12 on CPU as well) Robert Sachunsky 2025-09-30 19:16:50 +02:00
  • b21051db21 ProcessPoolExecutor: shutdown during del() instead of atexit() Robert Sachunsky 2025-09-30 19:16:00 +02:00
  • 08c8c26028 indent extremely long lines Robert Sachunsky 2025-09-30 03:52:19 +02:00
  • f857ee7b51 simplify Robert Sachunsky 2025-09-19 02:12:18 +02:00
  • c0137c29ad try to fix the failed outsourcing of utils_ocr Robert Sachunsky 2025-09-30 02:23:43 +02:00
  • 13f85b0d5c Merge branch 'main' into loky-with-shm-for-175-rebuilt Robert Sachunsky 2025-09-30 02:07:20 +02:00
  • 070dafca75 remove duplicate LICENSE cneud 2025-09-29 22:17:27 +02:00
  • 53c1ca11fc Update README.md cneud 2025-09-29 22:15:17 +02:00
  • 0d695a2e2d Install nvidia-cudnn11==8.6.0.163 as workaround to NHCW/NCHW issue, ht @bertsky workaround-nhcw-nchw-issue kba 2025-09-29 19:18:16 +02:00
  • 758602403e replace loky with concurrent.futures.ProcessPoolExecutor (faster) Robert Sachunsky 2025-09-21 21:35:22 +02:00
  • 0366707136 get_smallest_skew: do not pass logger Robert Sachunsky 2025-09-20 00:57:00 +02:00
  • b94c96fcbb find_num_col: exit early if empty (avoiding exceptions) Robert Sachunsky 2025-09-20 00:56:33 +02:00
  • 04c3d7dd1b get_smallest_skew: avoid shm if no ProcessPoolExecutor is passed Robert Sachunsky 2025-09-18 20:07:54 +02:00
  • 0662ece536 do_work_of_slopes*: use shm also in non-light mode(s) Robert Sachunsky 2025-09-04 15:18:55 +02:00
  • 31f240c3b8 do_image_rotation, do_work_of_slopes_new_curved: pass arrays via shared memory Robert Sachunsky 2025-09-02 15:04:04 +02:00
  • 8be2c79771 Revert "deskewing with faster multiprocessing" Robert Sachunsky 2025-09-03 09:01:18 +02:00
  • abf5c0f845 get_smallest_skew: when shifting search range of rotation angle, compare resulting (maximum) variances instead of blindly assuming the new range is better Robert Sachunsky 2025-09-02 15:01:52 +02:00
  • dc0caad512 writer: use @type='heading' instead of 'header' Robert Sachunsky 2025-08-26 21:07:50 +02:00
  • f458e3ece0 writer: SeparatorRegion needs SeparatorRegionType (not ImageRegionType) Robert Sachunsky 2025-08-26 21:07:18 +02:00
  • 4337d62985 contours: rename 'pixel' → 'label' for clarity Robert Sachunsky 2025-08-26 21:06:36 +02:00
  • 5b16c2fc00 avoid pulling unused 'image_page_rotated' through functions Robert Sachunsky 2025-08-26 21:05:40 +02:00
  • 5bff2d156a use box2rect instead of crop_image_inside_box when no image needed Robert Sachunsky 2025-08-26 21:02:43 +02:00
  • 9b5182c1c0 utils: introduce box2rect and box2slice Robert Sachunsky 2025-08-26 21:00:33 +02:00
  • bca2ae3d78 get_marginals: exit early if no peaks found to avoid spurious overlap mask Robert Sachunsky 2025-08-29 12:37:44 +02:00
  • 235539a350 filter_contours_without_textline_inside: avoid removing from identical lists twice Robert Sachunsky 2025-08-29 12:19:37 +02:00
  • 11e143afee polygon2contour: avoid overflow Robert Sachunsky 2025-08-29 12:16:56 +02:00
  • 7a9e8256ee increase dilatation: textregions/lines (5→6), seplines (0→1) Robert Sachunsky 2025-08-21 13:00:31 +02:00
  • f3faa29528 refactor shapely converisons into contour2polygon / polygon2contour, also handle heterogeneous geometries Robert Sachunsky 2025-08-21 12:59:03 +02:00
  • 0650274ffa move dilate_*_contours to .utils.contour, rename dilate_textregions_contours_textline_version → dilate_textline_contours Robert Sachunsky 2025-08-21 01:42:46 +02:00
  • a433c73628 filter_contours_area_of_image*: also ensure validity here Robert Sachunsky 2025-08-21 01:33:16 +02:00
  • 17bcf1af71 rename *lines_xml → *seplines for clarity Robert Sachunsky 2025-08-21 01:32:32 +02:00
  • e730725da3 check_any_text_region_in_model_one_is_main_or_header_light: return original instead of resampled contours Robert Sachunsky 2025-08-21 01:05:15 +02:00
  • 7b51fd6624 avoid creating invalid polygons via rounding Robert Sachunsky 2025-08-21 01:03:46 +02:00
  • 41cc38c51a get_textregion_contours_in_org_image_light: no back rotation, drop slope_first (always 0) Robert Sachunsky 2025-08-20 14:28:14 +02:00
  • afba70c920 separate_lines/do_work_of_slopes: skip if crop is empty Robert Sachunsky 2025-08-19 22:56:36 +02:00
  • 66b2bce8b9 return_boxes_of_images_by_order_of_reading_new: log any exceptions Robert Sachunsky 2025-09-19 12:19:58 +02:00
  • b48c41e68f return_boxes_of_images_by_order_of_reading_new: simplify, avoid changing dtype during np.append Robert Sachunsky 2025-08-19 20:09:09 +02:00
  • 09ece86f0d dilate_textregions_contours: simplify (via shapely's Polygon.buffer()), ensure validity Robert Sachunsky 2025-08-19 11:58:45 +02:00
  • 9d8b858dfc remove docs/eynollah-layout, superseded by docs/model.md and docs/usage.md kba 2025-09-29 16:01:29 +02:00
  • 2bcd20ebc7 reference the now-merged training tools in README.md kba 2025-09-29 15:21:42 +02:00
  • ce02a3553b 🔥 remove obsolete versions of the training document kba 2025-09-29 15:18:21 +02:00
  • 6d379782ab 📝 align former upstream train.md with wiki train.md syntactically kba 2025-09-29 15:11:02 +02:00
  • 52a7c93319 add documentation on training eynollah from sbb_pixelwise_segmentation wiki kba 2025-09-29 15:05:05 +02:00
  • ea05461dfe add documentation on eynollah layout from eynollah wiki kba 2025-09-29 15:04:46 +02:00
  • 56c4b7af88 📝 align pre-merge docs/train.md with former upstream train.md syntactically kba 2025-09-29 14:59:41 +02:00
  • 3b9548d0bd Merge sbb_pixelwise_segmentation training code into eynollah kba 2025-09-29 14:44:31 +02:00
  • a48e52c00e 📝 extend changelog for v0.5.0 Robert Sachunsky 2025-09-29 13:49:18 +02:00
  • a6f0af07d1
         Merge pull request #185 from bertsky/patch-4 Konstantin Baierer 2025-09-29 10:44:27 +02:00
  • 92c1e824dc
         CD: master is now main Robert Sachunsky 2025-09-26 23:05:47 +02:00
  • 6ea6a62801 📝 v0.5.0 v0.5.0 release-v0.5.0 kba 2025-09-26 16:23:46 +02:00
  • 882e242946
         Merge pull request #178 from qurator-spk/prepare-release-v0.5.0 Konstantin Baierer 2025-09-26 16:21:09 +02:00
  • 37e64b4e45 📝 changelog prepare-release-v0.5.0 kba 2025-09-26 16:19:04 +02:00
  • 3123add815 📝 update README kba 2025-09-26 15:07:32 +02:00
  • 830cc2c30a comment out the offending test outright kba 2025-09-26 14:37:04 +02:00
  • eb8d4573a8 tests: also disable ...ocr_directory test kba 2025-09-26 13:57:08 +02:00
  • 42fb452a7e disable the -doit OCR test kba 2025-09-26 12:54:29 +02:00
  • 480daa4c7c test_run: make ocr -doit work (add truetype file) Robert Sachunsky 2025-09-25 22:25:05 +02:00
  • 4c6405713a ci: ocr models kba 2025-09-25 22:19:36 +02:00
  • b4d460ca79 makefile forgot the OCR models adapt-ocrd kba 2025-09-25 22:16:38 +02:00
  • f3f5426597 Merge branch 'adapt-ocrd' of https://github.com/qurator-spk/eynollah into adapt-ocrd kba 2025-09-25 21:47:27 +02:00
  • 0bb1fb1a05 tests: adapt to layout/ocr model split kba 2025-09-25 21:47:15 +02:00
  • 2ec773128b Merge branch 'adapt-ocrd' of https://github.com/qurator-spk/eynollah into adapt-ocrd kba 2025-09-25 21:40:48 +02:00
  • f37d80c188 Merge branch 'adapt-ocrd' of https://github.com/qurator-spk/eynollah into adapt-ocrd kba 2025-09-25 21:39:55 +02:00
  • 57ee1cdc72 Merge remote-tracking branch 'bertsky/mbro_dead_code-plus-fixes-plus-tests' into adapt-ocrd kba 2025-09-25 21:39:36 +02:00
  • 5c0ab509c4 CI: Update model name kba 2025-09-25 21:17:32 +02:00
  • 9303ded11f ocrd-tool.json: use models_layout instead of eynollah_layouts for consistency kba 2025-09-25 21:12:52 +02:00
  • 7c79902835 enhancement/mbreorder: make all path options kwargs to run() instead of attributes Robert Sachunsky 2025-09-25 20:51:02 +02:00
  • e6ee26fde3 make models: adapt to zenodo/v0.5.0 kba 2025-09-25 20:35:54 +02:00
  • 85a3604a86
         Merge pull request #182 from bertsky/mbro_dead_code-plus-fixes-plus-tests mbro_dead_code Konstantin Baierer 2025-09-25 20:16:36 +02:00
  • 11de8a025d Adapt ocrd-eynollah-segment for release kba 2025-09-25 20:11:48 +02:00
  • 5e15c4f248 Merge remote-tracking branch 'bertsky/mbro_dead_code-plus-fixes-plus-tests' into prepare-release-v0.5.0 kba 2025-09-25 20:05:03 +02:00
  • 5c7e1f21fb test_run: add tests for ocr Robert Sachunsky 2025-09-25 19:53:19 +02:00
  • 2d14d57e4f ocr: minimal debug logging Robert Sachunsky 2025-09-25 19:52:50 +02:00
  • 1dcc7b5795 ocr CLI: make --model vs --model_name xor Robert Sachunsky 2025-09-25 16:38:43 +02:00
  • 5b1e0c1327 layout/ocr: make all path options kwargs to run() instead of attributes; ocr: drop redundant prediction_with_both_of_rgb_and_bin in favour of just bool(dir_in_bin) Robert Sachunsky 2025-09-25 16:26:31 +02:00
  • ef1304a764 CLIs: reorder options, explain -i vs -di Robert Sachunsky 2025-09-25 16:11:39 +02:00
  • df5448cdcd CLIs: add required=True where missing Robert Sachunsky 2025-09-25 16:08:40 +02:00
  • 58dd192fad smoke-test: also add enhancement and mbreorder here Robert Sachunsky 2025-09-25 16:05:45 +02:00
  • 369ef573f9 get textlines sorted in textregions - detection of vertical and horizontal regions improved b-vr103 2025-09-25 02:38:22 +02:00
  • be9ab60a5e get textlines sorted in textregions - detection of vertical and horizontal regions improved b-vr103 2025-09-25 02:38:22 +02:00
  • f07df080f0 add tests for enhancement and mbreorder Robert Sachunsky 2025-09-25 01:13:48 +02:00
  • 9967510327 mbreorder: filter by .xml suffix in dir-in mode Robert Sachunsky 2025-09-25 00:52:16 +02:00