1621532092
Merge branch 'main' into eynollah_light
2 years ago
4c217018cc
textline light version -tll can not work without enabling -light option
2 years ago
d68f240b59
loading TensorFlow SavedModel format is now present
2 years ago
380f59ad67
let hybrid textline light model be loaded
2 years ago
29e6ad076f
renaming textline light model
2 years ago
529f2c0e19
set_memory_growth to all GPU devices alike
2 years ago
8fe3567123
set_memory_growth to all GPU devices alike
2 years ago
d3735b12f4
pushing commits 2d9ccac and 7345f6b into eynollah_light
2 years ago
817e5a6af9
update docstring
2 years ago
a9728bb899
Update eynollah.py
...
predict quietly please
2 years ago
fb6d97091b
OCR-D wrapper: expose tables param
2 years ago
1ac0a7e06f
try loading as TF SavedModel instead of HDF5
2 years ago
73057d57d1
silentium!
2 years ago
3d54719c87
fix import
2 years ago
58ca226f2d
apply some fixes from main
2 years ago
6c8f60993f
Merge branch 'main' into 420-namespace-package
2 years ago
ea792d1e4a
📦 v0.2.0
2 years ago
8e894e5b7b
Merge remote-tracking branch 'bertsky/savedmodel-format'
2 years ago
7cd07dd550
use PEP420 style qurator namespace
2 years ago
71d0ec8dfe
📦 v0.1.0
2 years ago
875e4fe32b
log number of detected regions
2 years ago
318ea6acca
OCR-D wrapper: expose tables param
2 years ago
23f0c0b40a
ocrd-tool: replace by persistent model URL
2 years ago
5c26bdf402
ocrd-tool: add model archive to resmgr resources
2 years ago
7345f6bf67
remove TF1 session and GC controls, avoid repeating load_model
2 years ago
a56988a35a
contours: numpy now needs dtype=object
2 years ago
2d9ccac354
contours: simplify
2 years ago
ab4bb7cd7b
silentium!
2 years ago
79e897d3b2
try loading as TF SavedModel instead of HDF5
2 years 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
3 years ago
ae7c424889
Update eynollah.py
3 years ago
cd9920eea7
extracting page
3 years ago
735abc43f3
option to ignore page extraction
3 years ago
934bbd5892
cleanup
3 years ago
ecf117ca95
adapt to tf1.compat session mode in tf2
3 years ago
568391ec4a
require model command line option ( fix #59 ) ( #73 )
3 years ago
3bbbeecfec
all options are enabled for light version
3 years ago
Gerber, Mike
f27ac155ae
🧹 Downgrade "Patch size" log message to debug
...
Fixes gh-55.
3 years ago
adf10942fa
issue #55 resolved
3 years ago
2eacb9a8ec
renaming the models
3 years ago
c606391c31
flow from directory
3 years ago
cf5ef8f5ae
light version as option
3 years ago
b8a532180a
light version integration
3 years ago
10f1acef29
Merge pull request #65 from mikegerber/fix/enhanced-message
...
Fix/enhanced message
3 years ago
c30d4d5c30
Merge pull request #64 from mikegerber/feat/better-time-msgs
...
💄 Improve timing messages (Fixes #62 )
3 years ago
Gerber, Mike
11d9b00510
🧹 Don't produce spurious TextEquiv elements.
...
eynollah produces spurious - and empy - pcGts TextEquiv elements. This
is a. unnecessary, b. wrong and c. produces a lot of warning messages
in subsequent OCR processing steps because the OCR processor warns
about already existing text.
Fix this by not generating any TextEquiv elements.
Fixes gh-37.
3 years ago
vahidrezanezhad
vahid
Robert Sachunsky
cneud
Robert Sachunsky
Konstantin Baierer
cneud
Gerber, Mike