f09b7c1bef
use tf1 compatibility for keras backend
1 month ago
6018b354aa
comment unnecessary print commands
5 months ago
7983a65006
filtering separators in a correct way without missing them
7 months ago
7a70b20c77
apply missed commit #a56988a back
8 months ago
0ea90b7509
Merge pull request #116 from qurator-spk/fix-typos
...
fix some typos
9 months ago
d3b06baa84
fix some typos
9 months ago
935332e863
test fix for keras.backend import error with Python 3.8
9 months ago
5f7d9a475c
Merge pull request #109 from bertsky/patch-3
...
pil_cv2.check_dpi: fix class membership test
10 months ago
867a7261de
pil_cv2.check_dpi: fix class membership test
...
(depending on how the `PIL.Image` was instantiated – file plugin or array interface – the previous `isinstance` could fail, provoking a fall-through to `cv2pil` which does not work)
11 months ago
b049e475e3
update tool json resource path_in_archive
11 months ago
69c1d6b3d6
Revert "Merge pull request #97 from qurator-spk/420-namespace-package"
...
This reverts commit fd56b86acfea792d1e4a
11 months ago
0b35011847
right2left reading order detection accomplished
11 months ago
0cda1f3c7a
reading order type 1: right to left
11 months ago
b01888da31
delete printing resized image shape
12 months ago
45c40a58fc
issue #67 solved
12 months ago
6eab7a60a9
Update ocrd-tool.json
12 months ago
1621532092
Merge branch 'main' into eynollah_light
12 months ago
4c217018cc
textline light version -tll can not work without enabling -light option
1 year ago
d68f240b59
loading TensorFlow SavedModel format is now present
1 year ago
380f59ad67
let hybrid textline light model be loaded
1 year ago
29e6ad076f
renaming textline light model
1 year ago
529f2c0e19
set_memory_growth to all GPU devices alike
1 year ago
8fe3567123
set_memory_growth to all GPU devices alike
1 year ago
d3735b12f4
pushing commits 2d9ccac and 7345f6b into eynollah_light
1 year ago
817e5a6af9
update docstring
1 year ago
a9728bb899
Update eynollah.py
...
predict quietly please
1 year ago
fb6d97091b
OCR-D wrapper: expose tables param
1 year ago
1ac0a7e06f
try loading as TF SavedModel instead of HDF5
1 year ago
73057d57d1
silentium!
1 year ago
3d54719c87
fix import
1 year ago
58ca226f2d
apply some fixes from main
1 year ago
6c8f60993f
Merge branch 'main' into 420-namespace-package
1 year ago
ea792d1e4a
📦 v0.2.0
1 year ago
8e894e5b7b
Merge remote-tracking branch 'bertsky/savedmodel-format'
1 year ago
7cd07dd550
use PEP420 style qurator namespace
1 year ago
71d0ec8dfe
📦 v0.1.0
1 year ago
875e4fe32b
log number of detected regions
1 year ago
318ea6acca
OCR-D wrapper: expose tables param
1 year ago
23f0c0b40a
ocrd-tool: replace by persistent model URL
1 year ago
5c26bdf402
ocrd-tool: add model archive to resmgr resources
1 year ago
7345f6bf67
remove TF1 session and GC controls, avoid repeating load_model
1 year ago
a56988a35a
contours: numpy now needs dtype=object
1 year ago
2d9ccac354
contours: simplify
1 year ago
ab4bb7cd7b
silentium!
1 year ago
79e897d3b2
try loading as TF SavedModel instead of HDF5
1 year ago
38bf0d8740
solving issue by loading model by directory as input
2 years ago
583cdcee2c
new (hybrid cnn+transformer) textline model which can accelerate to extract contour textlines faster
2 years ago
8d5079c909
issue #77 is resolved on main branch
2 years ago
402c5339ac
issue #77 is resolved
2 years ago
01bfc3914d
extracting page as an option
2 years ago
cneud
vahidrezanezhad
Robert Sachunsky
Robert Sachunsky
vahid
Konstantin Baierer