1283 commits

This branch

This branch

All branches

Author	SHA1	Message	Date
vahidrezanezhad	c4434c7f7d	same task name for transformer-ocr training and inference	2026-02-19 13:59:16 +01:00
vahidrezanezhad	a84ae67e7a	fix a typo	2026-02-19 00:04:42 +01:00
vahidrezanezhad	77adcbea8a	copy characters list needed for cnn-rnn ocr model output while training and ensembling	2026-02-18 16:47:21 +01:00
vahidrezanezhad	4f66734e4d	eynollah config files has renamed from config.json to config_eynollah.json - training trocr model still misses to write config file into checkpoint directories	2026-02-18 16:04:44 +01:00
vahidrezanezhad	b426f7f152	trocr inference is integrated - works on CPU cause seg fault on GPU	2026-02-18 15:04:54 +01:00
vahidrezanezhad	733462381c	bug fix: layout visualization	2026-02-16 11:50:39 +01:00
vahidrezanezhad	9aa19aa6fa	cnn-rnn ocr inference: get input shape of model	2026-02-16 11:49:05 +01:00
vahidrezanezhad	68dd5eab62	fixing: imporing StringLookup	2026-02-12 15:37:11 +01:00
vahidrezanezhad	2ee8d8e050	Update inference.py ... importing StringLookup for cnn-rnn inference	2026-02-12 15:34:58 +01:00
vahidrezanezhad	ed034aa8ce	Update inference.py ... Fix broken inference model loading introduced during refactoring or merge	2026-02-12 15:28:15 +01:00
vahidrezanezhad	47fa22112c	import tensorflow is uncommented for ocr training	2026-02-11 19:52:56 +01:00
vahidrezanezhad	ab43477451	extracting ocr textline images and text: vertical lines threshold has changed to 1.4	2026-02-10 14:32:23 +01:00
vahidrezanezhad	a57914a68a	fixed: textline and text correct extraction for page xml if vertical textlines are excluded + textline and text extraction for page alto files	2026-02-09 18:53:08 +01:00
vahidrezanezhad	fbf252db13	torch model ensembling is integrated	2026-02-04 21:16:08 +01:00
vahidrezanezhad	498ff8f7a5	Updating the --help descriptions	2026-02-03 20:27:06 +01:00
vahidrezanezhad	fff4253352	generate or update list of characters in the case of cnn-rnn ocr training	2026-02-03 20:22:47 +01:00
vahidrezanezhad	60f0fb541d	integrating transformer ocr	2026-02-03 19:45:50 +01:00
kba	586077fbcd	📦 v0.7.0	2026-01-30 16:40:55 +01:00
kba	4ade0f788f	📝 changelog	2026-01-29 17:33:35 +01:00
kba	f13560726e	Merge remote-tracking branch 'origin/adding-cnn-rnn-training-script' into 2026-01-29-training ... # Conflicts: # src/eynollah/training/inference.py	2026-01-29 17:32:08 +01:00
kba	87190f8997	Merge branch 'adding-cnn-rnn-training-script-rfct' into 2026-01-29-training ... # Conflicts: # src/eynollah/training/models.py	2026-01-29 10:27:36 +01:00
kba	a76de1e182	Merge branch 'adding-cnn-rnn-training-script' into 2026-01-29-training	2026-01-29 10:26:34 +01:00
kba	ef3cf02877	Merge branch 'ruff-training' into 2026-01-29-training	2026-01-29 10:26:14 +01:00
vahidrezanezhad	f9695cd7be	Merge branch 'adding-cnn-rnn-training-script' of https://github.com/qurator-spk/eynollah into adding-cnn-rnn-training-script	2026-01-28 11:52:36 +01:00
vahidrezanezhad	3500167870	weights ensembling for tensorflow models is integrated	2026-01-28 11:52:12 +01:00
vahidrezanezhad	33f6a231bc	fix: prevent crash when printspace is missing in xmls used for label generation	2026-01-26 17:30:26 +01:00
vahidrezanezhad	6ae244bf9b	Fix filename stem extraction using binarization. Restore the CNN-RNN model to its previous version, as setting channels_last alone was insufficient for running on both CPU and GPU. Prevent errors caused by null values in image shape elements.	2026-01-26 15:04:47 +01:00
vahidrezanezhad	30f39e7383	mapregion is added to labels	2026-01-26 13:56:34 +01:00
vahidrezanezhad	c8240905a8	Fix label generation by selecting largest contour when erosion splits shapes	2026-01-26 13:36:24 +01:00
kba	9ccc495b4a	wip	2025-12-19 14:57:10 +01:00
vahidrezanezhad	49261fa99b	CNN–RNN–OCR inference and adaptation of the CNN–RNN–OCR model to support inference on both CPU and GPU	2025-12-17 15:12:39 +01:00
vahidrezanezhad	6ee79c7320	evaluation with a given GT is only possible for segmentation tasks	2025-12-17 13:28:02 +01:00
vahidrezanezhad	4651000191	debuging input shape + enable finetuning a model	2025-12-15 11:36:09 +01:00
vahidrezanezhad	4fc3ff33cb	The cnn-rnn ocr model can be trained now	2025-12-09 17:22:12 +01:00
vahidrezanezhad	84a72a128b	cnn-rnn model can be called - model input height and width are dynamic now - data generator is also callable	2025-12-09 15:30:19 +01:00
vahidrezanezhad	59e5a73654	adding cnn-rnn training script	2025-12-08 19:30:57 +01:00
vahidrezanezhad	7bf5e077d9	Restore correct execution of export_textline_images_and_text	2025-12-03 15:40:52 +01:00
vahidrezanezhad	6ac37af2f8	Fix eynollah ocr --help so it works again	2025-12-03 14:11:47 +01:00
vahidrezanezhad	d687d862d6	Restored correct functionality of the extract_only_images mode and cleaned up the argument handling	2025-12-03 12:01:42 +01:00
kba	51abe9617a	log to STDERR not STDOUT	2025-12-02 15:00:33 +01:00
kba	b161e33854	🔥 refactor eynollah ocr ... .	2025-11-28 15:45:21 +01:00
kba	30f9c695dc	move line-gt extraction out of ocr to eynollah-training	2025-11-28 15:12:31 +01:00
kba	951bd2fce6	CI: do not upgrade (now-unpineed) torch	2025-11-28 15:12:31 +01:00
kba	9bcfeab057	💀 remove dead code from eynollah.py	2025-11-28 12:52:28 +01:00
kba	5171e09c2d	eynollah.py: fix kwargs to writer	2025-11-28 12:52:28 +01:00
kba	c24cf94bce	enforce kwargs for writer.build_...	2025-11-28 12:52:28 +01:00
kba	4aa9543a7d	remove more branches after textline_light default true	2025-11-27 11:30:00 +01:00
kba	177d555ded	factor out extract_only_images as eynollah extract-images	2025-11-26 21:37:00 +01:00
kba	83e8b289da	🔥 drop light_version/textline_light (now default and implied)	2025-11-26 20:48:22 +01:00
kba	ca83cf934d	fix imports from src/cli/cli_*/*_cli	2025-11-26 20:48:14 +01:00