qurator-spk/eynollah
1
0
Fork 0
mirror of https://github.com/qurator-spk/eynollah.git synced 2025-06-09 04:09:54 +02:00
Code Issues Projects Releases Wiki Activity

merge main

Browse source
This commit is contained in:
kba 2025-01-20 14:45:27 +01:00
commit 869110f185
9 changed files with 513 additions and 270 deletions
Download patch file Download diff file Expand all files Collapse all files

2
.github/workflows/test-eynollah.yml vendored
View file

@ -34,3 +34,5 @@ jobs:
pip install -r requirements-test.txt pip install -r requirements-test.txt
- name: Test with pytest - name: Test with pytest
run: make test run: make test
- name: Test docker build
run: make docker

26
Dockerfile Normal file
View file

@ -0,0 +1,26 @@
ARG DOCKER_BASE_IMAGE
FROM $DOCKER_BASE_IMAGE
ARG VCS_REF
ARG BUILD_DATE
LABEL \
maintainer="https://ocr-d.de/kontakt" \
org.label-schema.vcs-ref=$VCS_REF \
org.label-schema.vcs-url="https://github.com/qurator-spk/eynollah" \
org.label-schema.build-date=$BUILD_DATE
ENV DEBIAN_FRONTEND=noninteractive
ENV PYTHONIOENCODING=utf8
ENV XDG_DATA_HOME=/usr/local/share
WORKDIR /build-eynollah
COPY src/ ./src
COPY pyproject.toml .
COPY requirements.txt .
COPY README.md .
COPY Makefile .
RUN apt-get install -y --no-install-recommends g++
RUN make install
WORKDIR /data
VOLUME /data

21
Makefile
View file

@ -1,6 +1,11 @@
EYNOLLAH_MODELS ?= $(PWD)/models_eynollah EYNOLLAH_MODELS ?= $(PWD)/models_eynollah
export EYNOLLAH_MODELS export EYNOLLAH_MODELS
# DOCKER_BASE_IMAGE = artefakt.dev.sbb.berlin:5000/sbb/ocrd_core:v2.68.0
DOCKER_BASE_IMAGE = docker.io/ocrd/core:v2.68.0
DOCKER_TAG = ocrd/eynollah
# BEGIN-EVAL makefile-parser --make-help Makefile # BEGIN-EVAL makefile-parser --make-help Makefile
help: help:
@ -22,17 +27,14 @@ help:
models: models_eynollah models: models_eynollah
models_eynollah: models_eynollah.tar.gz models_eynollah: models_eynollah.tar.gz
# tar xf models_eynollah_renamed.tar.gz --transform 's/models_eynollah_renamed/models_eynollah/'
# tar xf models_eynollah_renamed.tar.gz
# tar xf models_eynollah_renamed_savedmodel.tar.gz --transform 's/models_eynollah_renamed_savedmodel/models_eynollah/'
tar xf models_eynollah.tar.gz tar xf models_eynollah.tar.gz
models_eynollah.tar.gz: models_eynollah.tar.gz:
# wget 'https://qurator-data.de/eynollah/2021-04-25/models_eynollah.tar.gz' # wget 'https://qurator-data.de/eynollah/2021-04-25/models_eynollah.tar.gz'
# wget 'https://qurator-data.de/eynollah/2022-04-05/models_eynollah_renamed.tar.gz' # wget 'https://qurator-data.de/eynollah/2022-04-05/models_eynollah_renamed.tar.gz'
# wget 'https://ocr-d.kba.cloud/2022-04-05.SavedModel.tar.gz'
# wget 'https://qurator-data.de/eynollah/2022-04-05/models_eynollah_renamed_savedmodel.tar.gz' # wget 'https://qurator-data.de/eynollah/2022-04-05/models_eynollah_renamed_savedmodel.tar.gz'
wget https://github.com/qurator-spk/eynollah/releases/download/v0.3.0/models_eynollah.tar.gz # wget 'https://github.com/qurator-spk/eynollah/releases/download/v0.3.0/models_eynollah.tar.gz'
wget 'https://github.com/qurator-spk/eynollah/releases/download/v0.3.1/models_eynollah.tar.gz'
# Install with pip # Install with pip
install: install:
@ -48,3 +50,12 @@ smoke-test:
# Run unit tests # Run unit tests
test: test:
pytest tests pytest tests
# Build docker image
docker:
docker build \
--build-arg DOCKER_BASE_IMAGE=$(DOCKER_BASE_IMAGE) \
--build-arg VCS_REF=$$(git rev-parse --short HEAD) \
--build-arg BUILD_DATE=$$(date -u +"%Y-%m-%dT%H:%M:%SZ") \
-t $(DOCKER_TAG) .

3
README.md
View file

@ -21,7 +21,7 @@
:warning: Development is currently focused on achieving the best possible quality of results for a wide variety of historical documents and therefore processing can be very slow. We aim to improve this, but contributions are welcome. :warning: Development is currently focused on achieving the best possible quality of results for a wide variety of historical documents and therefore processing can be very slow. We aim to improve this, but contributions are welcome.
## Installation ## Installation
Python `3.8-3.11` with Tensorflow `2.12-2.15` on Linux are currently supported. Python `3.8-3.11` with Tensorflow `<2.13` on Linux are currently supported.
For (limited) GPU support the CUDA toolkit needs to be installed. For (limited) GPU support the CUDA toolkit needs to be installed.
@ -71,6 +71,7 @@ The following options can be used to further configure the processing:
| `-cl` | apply contour detection for curved text lines instead of bounding boxes | | `-cl` | apply contour detection for curved text lines instead of bounding boxes |
| `-ib` | apply binarization (the resulting image is saved to the output directory) | | `-ib` | apply binarization (the resulting image is saved to the output directory) |
| `-ep` | enable plotting (MUST always be used with `-sl`, `-sd`, `-sa`, `-si` or `-ae`) | | `-ep` | enable plotting (MUST always be used with `-sl`, `-sd`, `-sa`, `-si` or `-ae`) |
| `-eoi` | extract only images to output directory (other processing will not be done) |
| `-ho` | ignore headers for reading order dectection | | `-ho` | ignore headers for reading order dectection |
| `-si <directory>` | save image regions detected to this directory | | `-si <directory>` | save image regions detected to this directory |
| `-sd <directory>` | save deskewed image to this directory | | `-sd <directory>` | save deskewed image to this directory |

2
requirements.txt
View file

@ -2,7 +2,7 @@
ocrd >= 3.0.0b4 ocrd >= 3.0.0b4
numpy <1.24.0 numpy <1.24.0
scikit-learn >= 0.23.2 scikit-learn >= 0.23.2
tensorflow == 2.12.1 tensorflow < 2.13
imutils >= 0.5.3 imutils >= 0.5.3
matplotlib matplotlib
setuptools >= 50 setuptools >= 50

11
src/eynollah/cli.py
View file

@ -67,6 +67,12 @@ from eynollah.eynollah import Eynollah
is_flag=True, is_flag=True,
help="If set, will plot intermediary files and images", help="If set, will plot intermediary files and images",
) )
@click.option(
"--extract_only_images/--disable-extracting_only_images",
"-eoi/-noeoi",
is_flag=True,
help="If a directory is given, only images in documents will be cropped and saved there and the other processing will not be done",
)
@click.option( @click.option(
"--allow-enhancement/--no-allow-enhancement", "--allow-enhancement/--no-allow-enhancement",
"-ae/-noae", "-ae/-noae",
@ -148,6 +154,7 @@ def main(
save_layout, save_layout,
save_deskewed, save_deskewed,
save_all, save_all,
extract_only_images,
save_page, save_page,
enable_plotting, enable_plotting,
allow_enhancement, allow_enhancement,
@ -175,6 +182,9 @@ def main(
if textline_light and not light_version: if textline_light and not light_version:
print('Error: You used -tll to enable light textline detection but -light is not enabled') print('Error: You used -tll to enable light textline detection but -light is not enabled')
sys.exit(1) sys.exit(1)
if extract_only_images and (allow_enhancement or allow_scaling or light_version or curved_line or textline_light or full_layout or tables or right2left or headers_off) :
print('Error: You used -eoi which can not be enabled alongside light_version -light or allow_scaling -as or allow_enhancement -ae or curved_line -cl or textline_light -tll or full_layout -fl or tables -tab or right2left -r2l or headers_off -ho')
sys.exit(1)
eynollah = Eynollah( eynollah = Eynollah(
model, model,
getLogger('Eynollah'), getLogger('Eynollah'),
@ -182,6 +192,7 @@ def main(
dir_out=out, dir_out=out,
dir_in=dir_in, dir_in=dir_in,
dir_of_cropped_images=save_images, dir_of_cropped_images=save_images,
extract_only_images=extract_only_images,
dir_of_layout=save_layout, dir_of_layout=save_layout,
dir_of_deskewed=save_deskewed, dir_of_deskewed=save_deskewed,
dir_of_all=save_all, dir_of_all=save_all,

678
src/eynollah/eynollah.py
View file

@ -154,6 +154,7 @@ class Eynollah:
dir_out : Optional[str] = None, dir_out : Optional[str] = None,
dir_in : Optional[str] = None, dir_in : Optional[str] = None,
dir_of_cropped_images : Optional[str] = None, dir_of_cropped_images : Optional[str] = None,
extract_only_images : bool =False,
dir_of_layout : Optional[str] = None, dir_of_layout : Optional[str] = None,
dir_of_deskewed : Optional[str] = None, dir_of_deskewed : Optional[str] = None,
dir_of_all : Optional[str] = None, dir_of_all : Optional[str] = None,
@ -200,6 +201,7 @@ class Eynollah:
self.allow_scaling = allow_scaling self.allow_scaling = allow_scaling
self.headers_off = headers_off self.headers_off = headers_off
self.light_version = light_version self.light_version = light_version
self.extract_only_images = extract_only_images
self.ignore_page_extraction = ignore_page_extraction self.ignore_page_extraction = ignore_page_extraction
self.pcgts = pcgts self.pcgts = pcgts
if not dir_in: if not dir_in:
@ -230,6 +232,7 @@ class Eynollah:
self.model_page_dir = dir_models + "/eynollah-page-extraction_20210425" self.model_page_dir = dir_models + "/eynollah-page-extraction_20210425"
self.model_region_dir_p_ens = dir_models + "/eynollah-main-regions-ensembled_20210425" self.model_region_dir_p_ens = dir_models + "/eynollah-main-regions-ensembled_20210425"
self.model_region_dir_p_ens_light = dir_models + "/eynollah-main-regions_20220314" self.model_region_dir_p_ens_light = dir_models + "/eynollah-main-regions_20220314"
self.model_region_dir_p_ens_light_only_images_extraction = dir_models + "/eynollah-main-regions_20231127_672_org_ens_11_13_16_17_18"
if self.textline_light: if self.textline_light:
self.model_textline_dir = dir_models + "/eynollah-textline_light_20210425" self.model_textline_dir = dir_models + "/eynollah-textline_light_20210425"
else: else:
@ -254,7 +257,23 @@ class Eynollah:
self.ls_imgs = os.listdir(self.dir_in) self.ls_imgs = os.listdir(self.dir_in)
if dir_in and not light_version: if dir_in and self.extract_only_images:
config = tf.compat.v1.ConfigProto()
config.gpu_options.allow_growth = True
session = tf.compat.v1.Session(config=config)
set_session(session)
self.model_page = self.our_load_model(self.model_page_dir)
self.model_classifier = self.our_load_model(self.model_dir_of_col_classifier)
self.model_bin = self.our_load_model(self.model_dir_of_binarization)
#self.model_textline = self.our_load_model(self.model_textline_dir)
self.model_region = self.our_load_model(self.model_region_dir_p_ens_light_only_images_extraction)
#self.model_region_fl_np = self.our_load_model(self.model_region_dir_fully_np)
#self.model_region_fl = self.our_load_model(self.model_region_dir_fully)
self.ls_imgs = os.listdir(self.dir_in)
if dir_in and not (light_version or self.extract_only_images):
config = tf.compat.v1.ConfigProto() config = tf.compat.v1.ConfigProto()
config.gpu_options.allow_growth = True config.gpu_options.allow_growth = True
session = tf.compat.v1.Session(config=config) session = tf.compat.v1.Session(config=config)
@ -468,6 +487,27 @@ class Eynollah:
return img_new, num_column_is_classified return img_new, num_column_is_classified
def calculate_width_height_by_columns_extract_only_images(self, img, num_col, width_early, label_p_pred):
self.logger.debug("enter calculate_width_height_by_columns")
if num_col == 1:
img_w_new = 700
elif num_col == 2:
img_w_new = 900
elif num_col == 3:
img_w_new = 1500
elif num_col == 4:
img_w_new = 1800
elif num_col == 5:
img_w_new = 2200
elif num_col == 6:
img_w_new = 2500
img_h_new = int(img.shape[0] / float(img.shape[1]) * img_w_new)
img_new = resize_image(img, img_h_new, img_w_new)
num_column_is_classified = True
return img_new, num_column_is_classified
def resize_image_with_column_classifier(self, is_image_enhanced, img_bin): def resize_image_with_column_classifier(self, is_image_enhanced, img_bin):
self.logger.debug("enter resize_image_with_column_classifier") self.logger.debug("enter resize_image_with_column_classifier")
if self.input_binary: if self.input_binary:
@ -575,13 +615,18 @@ class Eynollah:
self.logger.info("Found %d columns (%s)", num_col, np.around(label_p_pred, decimals=5)) self.logger.info("Found %d columns (%s)", num_col, np.around(label_p_pred, decimals=5))
if dpi < DPI_THRESHOLD: if not self.extract_only_images:
img_new, num_column_is_classified = self.calculate_width_height_by_columns(img, num_col, width_early, label_p_pred) if dpi < DPI_THRESHOLD:
if light_version: img_new, num_column_is_classified = self.calculate_width_height_by_columns(img, num_col, width_early, label_p_pred)
image_res = np.copy(img_new) if light_version:
image_res = np.copy(img_new)
else:
image_res = self.predict_enhancement(img_new)
is_image_enhanced = True
else: else:
image_res = self.predict_enhancement(img_new) num_column_is_classified = True
is_image_enhanced = True image_res = np.copy(img)
is_image_enhanced = False
else: else:
num_column_is_classified = True num_column_is_classified = True
image_res = np.copy(img) image_res = np.copy(img)
@ -872,9 +917,13 @@ class Eynollah:
seg_not_base = label_p_pred[0,:,:,4] seg_not_base = label_p_pred[0,:,:,4]
##seg2 = -label_p_pred[0,:,:,2] ##seg2 = -label_p_pred[0,:,:,2]
if self.extract_only_images:
seg_not_base[seg_not_base>0.03] =1 #seg_not_base[seg_not_base>0.3] =1
seg_not_base[seg_not_base<1] =0 seg_not_base[seg_not_base>0.5] =1
seg_not_base[seg_not_base<1] =0
else:
seg_not_base[seg_not_base>0.03] =1
seg_not_base[seg_not_base<1] =0
@ -893,13 +942,10 @@ class Eynollah:
seg_line[seg_line>0.1] =1 seg_line[seg_line>0.1] =1
seg_line[seg_line<1] =0 seg_line[seg_line<1] =0
if not self.extract_only_images:
seg_background = label_p_pred[0,:,:,0] seg_background = label_p_pred[0,:,:,0]
##seg2 = -label_p_pred[0,:,:,2] seg_background[seg_background>0.25] =1
seg_background[seg_background<1] =0
seg_background[seg_background>0.25] =1
seg_background[seg_background<1] =0
##seg = seg+seg2 ##seg = seg+seg2
#seg = label_p_pred[0,:,:,2] #seg = label_p_pred[0,:,:,2]
#seg[seg>0.4] =1 #seg[seg>0.4] =1
@ -912,8 +958,9 @@ class Eynollah:
##plt.show() ##plt.show()
#seg[seg==1]=0 #seg[seg==1]=0
#seg[seg_test==1]=1 #seg[seg_test==1]=1
seg[seg_not_base==1]=4 ###seg[seg_not_base==1]=4
seg[seg_background==1]=0 if not self.extract_only_images:
seg[seg_background==1]=0
seg[(seg_line==1) & (seg==0)]=3 seg[(seg_line==1) & (seg==0)]=3
seg_color = np.repeat(seg[:, :, np.newaxis], 3, axis=2) seg_color = np.repeat(seg[:, :, np.newaxis], 3, axis=2)
@ -1578,6 +1625,124 @@ class Eynollah:
q.put(slopes_sub) q.put(slopes_sub)
poly.put(poly_sub) poly.put(poly_sub)
box_sub.put(boxes_sub_new) box_sub.put(boxes_sub_new)
def get_regions_light_v_extract_only_images(self,img,is_image_enhanced, num_col_classifier):
self.logger.debug("enter get_regions_extract_images_only")
erosion_hurts = False
img_org = np.copy(img)
img_height_h = img_org.shape[0]
img_width_h = img_org.shape[1]
if num_col_classifier == 1:
img_w_new = 700
elif num_col_classifier == 2:
img_w_new = 900
elif num_col_classifier == 3:
img_w_new = 1500
elif num_col_classifier == 4:
img_w_new = 1800
elif num_col_classifier == 5:
img_w_new = 2200
elif num_col_classifier == 6:
img_w_new = 2500
img_h_new = int(img.shape[0] / float(img.shape[1]) * img_w_new)
img_resized = resize_image(img,img_h_new, img_w_new )
if not self.dir_in:
model_region, session_region = self.start_new_session_and_model(self.model_region_dir_p_ens_light_only_images_extraction)
prediction_regions_org = self.do_prediction_new_concept(True, img_resized, model_region)
else:
prediction_regions_org = self.do_prediction_new_concept(True, img_resized, self.model_region)
#plt.imshow(prediction_regions_org[:,:,0])
#plt.show()
prediction_regions_org = resize_image(prediction_regions_org,img_height_h, img_width_h )
image_page, page_coord, cont_page = self.extract_page()
prediction_regions_org = prediction_regions_org[page_coord[0] : page_coord[1], page_coord[2] : page_coord[3]]
prediction_regions_org=prediction_regions_org[:,:,0]
mask_lines_only = (prediction_regions_org[:,:] ==3)*1
mask_texts_only = (prediction_regions_org[:,:] ==1)*1
mask_images_only=(prediction_regions_org[:,:] ==2)*1
polygons_lines_xml, hir_lines_xml = return_contours_of_image(mask_lines_only)
polygons_lines_xml = textline_con_fil = filter_contours_area_of_image(mask_lines_only, polygons_lines_xml, hir_lines_xml, max_area=1, min_area=0.00001)
polygons_of_only_texts = return_contours_of_interested_region(mask_texts_only,1,0.00001)
polygons_of_only_lines = return_contours_of_interested_region(mask_lines_only,1,0.00001)
text_regions_p_true = np.zeros(prediction_regions_org.shape)
text_regions_p_true = cv2.fillPoly(text_regions_p_true, pts = polygons_of_only_lines, color=(3,3,3))
text_regions_p_true[:,:][mask_images_only[:,:] == 1] = 2
text_regions_p_true = cv2.fillPoly(text_regions_p_true, pts = polygons_of_only_texts, color=(1,1,1))
text_regions_p_true[text_regions_p_true.shape[0]-15:text_regions_p_true.shape[0], :] = 0
text_regions_p_true[:, text_regions_p_true.shape[1]-15:text_regions_p_true.shape[1]] = 0
##polygons_of_images = return_contours_of_interested_region(text_regions_p_true, 2, 0.0001)
polygons_of_images = return_contours_of_interested_region(text_regions_p_true, 2, 0.001)
image_boundary_of_doc = np.zeros((text_regions_p_true.shape[0], text_regions_p_true.shape[1]))
###image_boundary_of_doc[:6, :] = 1
###image_boundary_of_doc[text_regions_p_true.shape[0]-6:text_regions_p_true.shape[0], :] = 1
###image_boundary_of_doc[:, :6] = 1
###image_boundary_of_doc[:, text_regions_p_true.shape[1]-6:text_regions_p_true.shape[1]] = 1
#plt.imshow(image_boundary_of_doc)
#plt.show()
polygons_of_images_fin = []
for ploy_img_ind in polygons_of_images:
"""
test_poly_image = np.zeros((text_regions_p_true.shape[0], text_regions_p_true.shape[1]))
test_poly_image = cv2.fillPoly(test_poly_image, pts = [ploy_img_ind], color=(1,1,1))
test_poly_image = test_poly_image[:,:] + image_boundary_of_doc[:,:]
test_poly_image_intersected_area = ( test_poly_image[:,:]==2 )*1
test_poly_image_intersected_area = test_poly_image_intersected_area.sum()
if test_poly_image_intersected_area==0:
##polygons_of_images_fin.append(ploy_img_ind)
x, y, w, h = cv2.boundingRect(ploy_img_ind)
box = [x, y, w, h]
_, page_coord_img = crop_image_inside_box(box, text_regions_p_true)
#cont_page.append(np.array([[page_coord[2], page_coord[0]], [page_coord[3], page_coord[0]], [page_coord[3], page_coord[1]], [page_coord[2], page_coord[1]]]))
polygons_of_images_fin.append(np.array([[page_coord_img[2], page_coord_img[0]], [page_coord_img[3], page_coord_img[0]], [page_coord_img[3], page_coord_img[1]], [page_coord_img[2], page_coord_img[1]]]) )
"""
x, y, w, h = cv2.boundingRect(ploy_img_ind)
if h < 150 or w < 150:
pass
else:
box = [x, y, w, h]
_, page_coord_img = crop_image_inside_box(box, text_regions_p_true)
#cont_page.append(np.array([[page_coord[2], page_coord[0]], [page_coord[3], page_coord[0]], [page_coord[3], page_coord[1]], [page_coord[2], page_coord[1]]]))
polygons_of_images_fin.append(np.array([[page_coord_img[2], page_coord_img[0]], [page_coord_img[3], page_coord_img[0]], [page_coord_img[3], page_coord_img[1]], [page_coord_img[2], page_coord_img[1]]]) )
return text_regions_p_true, erosion_hurts, polygons_lines_xml, polygons_of_images_fin, image_page, page_coord, cont_page
def get_regions_light_v(self,img,is_image_enhanced, num_col_classifier): def get_regions_light_v(self,img,is_image_enhanced, num_col_classifier):
self.logger.debug("enter get_regions_light_v") self.logger.debug("enter get_regions_light_v")
erosion_hurts = False erosion_hurts = False
@ -2429,6 +2594,7 @@ class Eynollah:
prediction_table_erode = cv2.erode(prediction_table[:,:,0], KERNEL, iterations=20) prediction_table_erode = cv2.erode(prediction_table[:,:,0], KERNEL, iterations=20)
prediction_table_erode = cv2.dilate(prediction_table_erode, KERNEL, iterations=20) prediction_table_erode = cv2.dilate(prediction_table_erode, KERNEL, iterations=20)
return prediction_table_erode.astype(np.int16) return prediction_table_erode.astype(np.int16)
def run_graphics_and_columns_light(self, text_regions_p_1, textline_mask_tot_ea, num_col_classifier, num_column_is_classified, erosion_hurts): def run_graphics_and_columns_light(self, text_regions_p_1, textline_mask_tot_ea, num_col_classifier, num_column_is_classified, erosion_hurts):
img_g = self.imread(grayscale=True, uint8=True) img_g = self.imread(grayscale=True, uint8=True)
@ -2830,7 +2996,6 @@ class Eynollah:
""" """
self.logger.debug("enter run") self.logger.debug("enter run")
t0_tot = time.time() t0_tot = time.time()
if not self.dir_in: if not self.dir_in:
@ -2841,276 +3006,295 @@ class Eynollah:
if self.dir_in: if self.dir_in:
self.reset_file_name_dir(os.path.join(self.dir_in,img_name)) self.reset_file_name_dir(os.path.join(self.dir_in,img_name))
img_res, is_image_enhanced, num_col_classifier, num_column_is_classified = self.run_enhancement(self.light_version)
self.logger.info("Enhancing took %.1fs ", time.time() - t0)
t1 = time.time() if self.extract_only_images:
if self.light_version: img_res, is_image_enhanced, num_col_classifier, num_column_is_classified = self.run_enhancement(self.light_version)
text_regions_p_1 ,erosion_hurts, polygons_lines_xml, textline_mask_tot_ea = self.get_regions_light_v(img_res, is_image_enhanced, num_col_classifier) self.logger.info("Enhancing took %.1fs ", time.time() - t0)
slope_deskew, slope_first = self.run_deskew(textline_mask_tot_ea)
#self.logger.info("Textregion detection took %.1fs ", time.time() - t1t)
num_col, num_col_classifier, img_only_regions, page_coord, image_page, mask_images, mask_lines, text_regions_p_1, cont_page, table_prediction, textline_mask_tot_ea = \
self.run_graphics_and_columns_light(text_regions_p_1, textline_mask_tot_ea, num_col_classifier, num_column_is_classified, erosion_hurts)
#self.logger.info("run graphics %.1fs ", time.time() - t1t)
textline_mask_tot_ea_org = np.copy(textline_mask_tot_ea)
else:
text_regions_p_1 ,erosion_hurts, polygons_lines_xml = self.get_regions_from_xy_2models(img_res, is_image_enhanced, num_col_classifier)
self.logger.info("Textregion detection took %.1fs ", time.time() - t1)
t1 = time.time() text_regions_p_1 ,erosion_hurts, polygons_lines_xml,polygons_of_images,image_page, page_coord, cont_page = self.get_regions_light_v_extract_only_images(img_res, is_image_enhanced, num_col_classifier)
num_col, num_col_classifier, img_only_regions, page_coord, image_page, mask_images, mask_lines, text_regions_p_1, cont_page, table_prediction = \
self.run_graphics_and_columns(text_regions_p_1, num_col_classifier, num_column_is_classified, erosion_hurts) pcgts = self.writer.build_pagexml_no_full_layout([], page_coord, [], [], [], [], polygons_of_images, [], [], [], [], [], cont_page, [], [])
self.logger.info("Graphics detection took %.1fs ", time.time() - t1)
#self.logger.info('cont_page %s', cont_page) if self.plotter:
self.plotter.write_images_into_directory(polygons_of_images, image_page)
if not num_col:
self.logger.info("No columns detected, outputting an empty PAGE-XML")
pcgts = self.writer.build_pagexml_no_full_layout([], page_coord, [], [], [], [], [], [], [], [], [], [], cont_page, [], [])
self.logger.info("Job done in %.1fs", time.time() - t1)
if self.dir_in: if self.dir_in:
self.writer.write_pagexml(pcgts) self.writer.write_pagexml(pcgts)
continue
else: else:
return pcgts return pcgts
t1 = time.time() else:
if not self.light_version: img_res, is_image_enhanced, num_col_classifier, num_column_is_classified = self.run_enhancement(self.light_version)
textline_mask_tot_ea = self.run_textline(image_page) self.logger.info("Enhancing took %.1fs ", time.time() - t0)
self.logger.info("textline detection took %.1fs", time.time() - t1)
t1 = time.time() t1 = time.time()
slope_deskew, slope_first = self.run_deskew(textline_mask_tot_ea) if self.light_version:
self.logger.info("deskewing took %.1fs", time.time() - t1) text_regions_p_1 ,erosion_hurts, polygons_lines_xml, textline_mask_tot_ea = self.get_regions_light_v(img_res, is_image_enhanced, num_col_classifier)
t1 = time.time() slope_deskew, slope_first = self.run_deskew(textline_mask_tot_ea)
#plt.imshow(table_prediction) #self.logger.info("Textregion detection took %.1fs ", time.time() - t1t)
#plt.show() num_col, num_col_classifier, img_only_regions, page_coord, image_page, mask_images, mask_lines, text_regions_p_1, cont_page, table_prediction, textline_mask_tot_ea = \
self.run_graphics_and_columns_light(text_regions_p_1, textline_mask_tot_ea, num_col_classifier, num_column_is_classified, erosion_hurts)
#self.logger.info("run graphics %.1fs ", time.time() - t1t)
textline_mask_tot_ea_org = np.copy(textline_mask_tot_ea)
else:
text_regions_p_1 ,erosion_hurts, polygons_lines_xml = self.get_regions_from_xy_2models(img_res, is_image_enhanced, num_col_classifier)
self.logger.info("Textregion detection took %.1fs ", time.time() - t1)
textline_mask_tot, text_regions_p, image_page_rotated = self.run_marginals(image_page, textline_mask_tot_ea, mask_images, mask_lines, num_col_classifier, slope_deskew, text_regions_p_1, table_prediction) t1 = time.time()
self.logger.info("detection of marginals took %.1fs", time.time() - t1) num_col, num_col_classifier, img_only_regions, page_coord, image_page, mask_images, mask_lines, text_regions_p_1, cont_page, table_prediction = \
t1 = time.time() self.run_graphics_and_columns(text_regions_p_1, num_col_classifier, num_column_is_classified, erosion_hurts)
if not self.full_layout: self.logger.info("Graphics detection took %.1fs ", time.time() - t1)
polygons_of_images, img_revised_tab, text_regions_p_1_n, textline_mask_tot_d, regions_without_separators_d, boxes, boxes_d, polygons_of_marginals, contours_tables = self.run_boxes_no_full_layout(image_page, textline_mask_tot, text_regions_p, slope_deskew, num_col_classifier, table_prediction, erosion_hurts) #self.logger.info('cont_page %s', cont_page)
if self.full_layout: if not num_col:
polygons_of_images, img_revised_tab, text_regions_p_1_n, textline_mask_tot_d, regions_without_separators_d, regions_fully, regions_without_separators, polygons_of_marginals, contours_tables = self.run_boxes_full_layout(image_page, textline_mask_tot, text_regions_p, slope_deskew, num_col_classifier, img_only_regions, table_prediction, erosion_hurts) self.logger.info("No columns detected, outputting an empty PAGE-XML")
text_only = ((img_revised_tab[:, :] == 1)) * 1 pcgts = self.writer.build_pagexml_no_full_layout([], page_coord, [], [], [], [], [], [], [], [], [], [], cont_page, [], [])
if np.abs(slope_deskew) >= SLOPE_THRESHOLD: self.logger.info("Job done in %.1fs", time.time() - t1)
text_only_d = ((text_regions_p_1_n[:, :] == 1)) * 1 if self.dir_in:
self.writer.write_pagexml(pcgts)
continue
else:
return pcgts
t1 = time.time()
if not self.light_version:
textline_mask_tot_ea = self.run_textline(image_page)
self.logger.info("textline detection took %.1fs", time.time() - t1)
t1 = time.time()
slope_deskew, slope_first = self.run_deskew(textline_mask_tot_ea)
self.logger.info("deskewing took %.1fs", time.time() - t1)
t1 = time.time()
#plt.imshow(table_prediction)
#plt.show()
textline_mask_tot, text_regions_p, image_page_rotated = self.run_marginals(image_page, textline_mask_tot_ea, mask_images, mask_lines, num_col_classifier, slope_deskew, text_regions_p_1, table_prediction)
self.logger.info("detection of marginals took %.1fs", time.time() - t1)
t1 = time.time()
if not self.full_layout:
polygons_of_images, img_revised_tab, text_regions_p_1_n, textline_mask_tot_d, regions_without_separators_d, boxes, boxes_d, polygons_of_marginals, contours_tables = self.run_boxes_no_full_layout(image_page, textline_mask_tot, text_regions_p, slope_deskew, num_col_classifier, table_prediction, erosion_hurts)
if self.full_layout:
polygons_of_images, img_revised_tab, text_regions_p_1_n, textline_mask_tot_d, regions_without_separators_d, regions_fully, regions_without_separators, polygons_of_marginals, contours_tables = self.run_boxes_full_layout(image_page, textline_mask_tot, text_regions_p, slope_deskew, num_col_classifier, img_only_regions, table_prediction, erosion_hurts)
text_only = ((img_revised_tab[:, :] == 1)) * 1
if np.abs(slope_deskew) >= SLOPE_THRESHOLD:
text_only_d = ((text_regions_p_1_n[:, :] == 1)) * 1
min_con_area = 0.000005 min_con_area = 0.000005
if np.abs(slope_deskew) >= SLOPE_THRESHOLD: if np.abs(slope_deskew) >= SLOPE_THRESHOLD:
contours_only_text, hir_on_text = return_contours_of_image(text_only) contours_only_text, hir_on_text = return_contours_of_image(text_only)
contours_only_text_parent = return_parent_contours(contours_only_text, hir_on_text) contours_only_text_parent = return_parent_contours(contours_only_text, hir_on_text)
if len(contours_only_text_parent) > 0: if len(contours_only_text_parent) > 0:
areas_cnt_text = np.array([cv2.contourArea(c) for c in contours_only_text_parent]) areas_cnt_text = np.array([cv2.contourArea(c) for c in contours_only_text_parent])
areas_cnt_text = areas_cnt_text / float(text_only.shape[0] * text_only.shape[1]) areas_cnt_text = areas_cnt_text / float(text_only.shape[0] * text_only.shape[1])
#self.logger.info('areas_cnt_text %s', areas_cnt_text) #self.logger.info('areas_cnt_text %s', areas_cnt_text)
contours_biggest = contours_only_text_parent[np.argmax(areas_cnt_text)] contours_biggest = contours_only_text_parent[np.argmax(areas_cnt_text)]
contours_only_text_parent = [c for jz, c in enumerate(contours_only_text_parent) if areas_cnt_text[jz] > min_con_area] contours_only_text_parent = [c for jz, c in enumerate(contours_only_text_parent) if areas_cnt_text[jz] > min_con_area]
areas_cnt_text_parent = [area for area in areas_cnt_text if area > min_con_area] areas_cnt_text_parent = [area for area in areas_cnt_text if area > min_con_area]
index_con_parents = np.argsort(areas_cnt_text_parent) index_con_parents = np.argsort(areas_cnt_text_parent)
contours_only_text_parent = list(np.array(contours_only_text_parent,dtype=object)[index_con_parents]) contours_only_text_parent = list(np.array(contours_only_text_parent,dtype=object)[index_con_parents])
areas_cnt_text_parent = list(np.array(areas_cnt_text_parent)[index_con_parents]) areas_cnt_text_parent = list(np.array(areas_cnt_text_parent)[index_con_parents])
cx_bigest_big, cy_biggest_big, _, _, _, _, _ = find_new_features_of_contours([contours_biggest]) cx_bigest_big, cy_biggest_big, _, _, _, _, _ = find_new_features_of_contours([contours_biggest])
cx_bigest, cy_biggest, _, _, _, _, _ = find_new_features_of_contours(contours_only_text_parent) cx_bigest, cy_biggest, _, _, _, _, _ = find_new_features_of_contours(contours_only_text_parent)
contours_only_text_d, hir_on_text_d = return_contours_of_image(text_only_d) contours_only_text_d, hir_on_text_d = return_contours_of_image(text_only_d)
contours_only_text_parent_d = return_parent_contours(contours_only_text_d, hir_on_text_d) contours_only_text_parent_d = return_parent_contours(contours_only_text_d, hir_on_text_d)
areas_cnt_text_d = np.array([cv2.contourArea(c) for c in contours_only_text_parent_d]) areas_cnt_text_d = np.array([cv2.contourArea(c) for c in contours_only_text_parent_d])
areas_cnt_text_d = areas_cnt_text_d / float(text_only_d.shape[0] * text_only_d.shape[1]) areas_cnt_text_d = areas_cnt_text_d / float(text_only_d.shape[0] * text_only_d.shape[1])
if len(areas_cnt_text_d)>0: if len(areas_cnt_text_d)>0:
contours_biggest_d = contours_only_text_parent_d[np.argmax(areas_cnt_text_d)] contours_biggest_d = contours_only_text_parent_d[np.argmax(areas_cnt_text_d)]
index_con_parents_d = np.argsort(areas_cnt_text_d) index_con_parents_d = np.argsort(areas_cnt_text_d)
contours_only_text_parent_d = list(np.array(contours_only_text_parent_d,dtype=object)[index_con_parents_d]) contours_only_text_parent_d = list(np.array(contours_only_text_parent_d,dtype=object)[index_con_parents_d])
areas_cnt_text_d = list(np.array(areas_cnt_text_d)[index_con_parents_d]) areas_cnt_text_d = list(np.array(areas_cnt_text_d)[index_con_parents_d])
cx_bigest_d_big, cy_biggest_d_big, _, _, _, _, _ = find_new_features_of_contours([contours_biggest_d]) cx_bigest_d_big, cy_biggest_d_big, _, _, _, _, _ = find_new_features_of_contours([contours_biggest_d])
cx_bigest_d, cy_biggest_d, _, _, _, _, _ = find_new_features_of_contours(contours_only_text_parent_d) cx_bigest_d, cy_biggest_d, _, _, _, _, _ = find_new_features_of_contours(contours_only_text_parent_d)
try: try:
if len(cx_bigest_d) >= 5: if len(cx_bigest_d) >= 5:
cx_bigest_d_last5 = cx_bigest_d[-5:] cx_bigest_d_last5 = cx_bigest_d[-5:]
cy_biggest_d_last5 = cy_biggest_d[-5:] cy_biggest_d_last5 = cy_biggest_d[-5:]
dists_d = [math.sqrt((cx_bigest_big[0] - cx_bigest_d_last5[j]) ** 2 + (cy_biggest_big[0] - cy_biggest_d_last5[j]) ** 2) for j in range(len(cy_biggest_d_last5))] dists_d = [math.sqrt((cx_bigest_big[0] - cx_bigest_d_last5[j]) ** 2 + (cy_biggest_big[0] - cy_biggest_d_last5[j]) ** 2) for j in range(len(cy_biggest_d_last5))]
ind_largest = len(cx_bigest_d) -5 + np.argmin(dists_d) ind_largest = len(cx_bigest_d) -5 + np.argmin(dists_d)
else: else:
cx_bigest_d_last5 = cx_bigest_d[-len(cx_bigest_d):] cx_bigest_d_last5 = cx_bigest_d[-len(cx_bigest_d):]
cy_biggest_d_last5 = cy_biggest_d[-len(cx_bigest_d):] cy_biggest_d_last5 = cy_biggest_d[-len(cx_bigest_d):]
dists_d = [math.sqrt((cx_bigest_big[0]-cx_bigest_d_last5[j])**2 + (cy_biggest_big[0]-cy_biggest_d_last5[j])**2) for j in range(len(cy_biggest_d_last5))] dists_d = [math.sqrt((cx_bigest_big[0]-cx_bigest_d_last5[j])**2 + (cy_biggest_big[0]-cy_biggest_d_last5[j])**2) for j in range(len(cy_biggest_d_last5))]
ind_largest = len(cx_bigest_d) - len(cx_bigest_d) + np.argmin(dists_d) ind_largest = len(cx_bigest_d) - len(cx_bigest_d) + np.argmin(dists_d)
cx_bigest_d_big[0] = cx_bigest_d[ind_largest] cx_bigest_d_big[0] = cx_bigest_d[ind_largest]
cy_biggest_d_big[0] = cy_biggest_d[ind_largest] cy_biggest_d_big[0] = cy_biggest_d[ind_largest]
except Exception as why: except Exception as why:
self.logger.error(why) self.logger.error(why)
(h, w) = text_only.shape[:2] (h, w) = text_only.shape[:2]
center = (w // 2.0, h // 2.0) center = (w // 2.0, h // 2.0)
M = cv2.getRotationMatrix2D(center, slope_deskew, 1.0) M = cv2.getRotationMatrix2D(center, slope_deskew, 1.0)
M_22 = np.array(M)[:2, :2] M_22 = np.array(M)[:2, :2]
p_big = np.dot(M_22, [cx_bigest_big, cy_biggest_big]) p_big = np.dot(M_22, [cx_bigest_big, cy_biggest_big])
x_diff = p_big[0] - cx_bigest_d_big x_diff = p_big[0] - cx_bigest_d_big
y_diff = p_big[1] - cy_biggest_d_big y_diff = p_big[1] - cy_biggest_d_big
contours_only_text_parent_d_ordered = []
for i in range(len(contours_only_text_parent)):
p = np.dot(M_22, [cx_bigest[i], cy_biggest[i]])
p[0] = p[0] - x_diff[0]
p[1] = p[1] - y_diff[0]
dists = [math.sqrt((p[0] - cx_bigest_d[j]) ** 2 + (p[1] - cy_biggest_d[j]) ** 2) for j in range(len(cx_bigest_d))]
contours_only_text_parent_d_ordered.append(contours_only_text_parent_d[np.argmin(dists)])
# img2=np.zeros((text_only.shape[0],text_only.shape[1],3))
# img2=cv2.fillPoly(img2,pts=[contours_only_text_parent_d[np.argmin(dists)]] ,color=(1,1,1))
# plt.imshow(img2[:,:,0])
# plt.show()
else:
contours_only_text_parent_d_ordered = []
contours_only_text_parent_d = []
contours_only_text_parent = []
contours_only_text_parent_d_ordered = []
for i in range(len(contours_only_text_parent)):
p = np.dot(M_22, [cx_bigest[i], cy_biggest[i]])
p[0] = p[0] - x_diff[0]
p[1] = p[1] - y_diff[0]
dists = [math.sqrt((p[0] - cx_bigest_d[j]) ** 2 + (p[1] - cy_biggest_d[j]) ** 2) for j in range(len(cx_bigest_d))]
contours_only_text_parent_d_ordered.append(contours_only_text_parent_d[np.argmin(dists)])
# img2=np.zeros((text_only.shape[0],text_only.shape[1],3))
# img2=cv2.fillPoly(img2,pts=[contours_only_text_parent_d[np.argmin(dists)]] ,color=(1,1,1))
# plt.imshow(img2[:,:,0])
# plt.show()
else: else:
contours_only_text_parent_d_ordered = [] contours_only_text_parent_d_ordered = []
contours_only_text_parent_d = [] contours_only_text_parent_d = []
contours_only_text_parent = [] contours_only_text_parent = []
else: else:
contours_only_text_parent_d_ordered = [] contours_only_text, hir_on_text = return_contours_of_image(text_only)
contours_only_text_parent_d = [] contours_only_text_parent = return_parent_contours(contours_only_text, hir_on_text)
contours_only_text_parent = []
else:
contours_only_text, hir_on_text = return_contours_of_image(text_only)
contours_only_text_parent = return_parent_contours(contours_only_text, hir_on_text)
if len(contours_only_text_parent) > 0: if len(contours_only_text_parent) > 0:
areas_cnt_text = np.array([cv2.contourArea(c) for c in contours_only_text_parent]) areas_cnt_text = np.array([cv2.contourArea(c) for c in contours_only_text_parent])
areas_cnt_text = areas_cnt_text / float(text_only.shape[0] * text_only.shape[1]) areas_cnt_text = areas_cnt_text / float(text_only.shape[0] * text_only.shape[1])
contours_biggest = contours_only_text_parent[np.argmax(areas_cnt_text)] contours_biggest = contours_only_text_parent[np.argmax(areas_cnt_text)]
contours_only_text_parent = [c for jz, c in enumerate(contours_only_text_parent) if areas_cnt_text[jz] > min_con_area] contours_only_text_parent = [c for jz, c in enumerate(contours_only_text_parent) if areas_cnt_text[jz] > min_con_area]
areas_cnt_text_parent = [area for area in areas_cnt_text if area > min_con_area] areas_cnt_text_parent = [area for area in areas_cnt_text if area > min_con_area]
index_con_parents = np.argsort(areas_cnt_text_parent) index_con_parents = np.argsort(areas_cnt_text_parent)
contours_only_text_parent = list(np.array(contours_only_text_parent,dtype=object)[index_con_parents]) contours_only_text_parent = list(np.array(contours_only_text_parent,dtype=object)[index_con_parents])
areas_cnt_text_parent = list(np.array(areas_cnt_text_parent)[index_con_parents]) areas_cnt_text_parent = list(np.array(areas_cnt_text_parent)[index_con_parents])
cx_bigest_big, cy_biggest_big, _, _, _, _, _ = find_new_features_of_contours([contours_biggest]) cx_bigest_big, cy_biggest_big, _, _, _, _, _ = find_new_features_of_contours([contours_biggest])
cx_bigest, cy_biggest, _, _, _, _, _ = find_new_features_of_contours(contours_only_text_parent) cx_bigest, cy_biggest, _, _, _, _, _ = find_new_features_of_contours(contours_only_text_parent)
#self.logger.debug('areas_cnt_text_parent %s', areas_cnt_text_parent) #self.logger.debug('areas_cnt_text_parent %s', areas_cnt_text_parent)
# self.logger.debug('areas_cnt_text_parent_d %s', areas_cnt_text_parent_d) # self.logger.debug('areas_cnt_text_parent_d %s', areas_cnt_text_parent_d)
# self.logger.debug('len(contours_only_text_parent) %s', len(contours_only_text_parent_d)) # self.logger.debug('len(contours_only_text_parent) %s', len(contours_only_text_parent_d))
else: else:
pass pass
if self.light_version:
txt_con_org = get_textregion_contours_in_org_image_light(contours_only_text_parent, self.image, slope_first)
else:
txt_con_org = get_textregion_contours_in_org_image(contours_only_text_parent, self.image, slope_first)
boxes_text, _ = get_text_region_boxes_by_given_contours(contours_only_text_parent)
boxes_marginals, _ = get_text_region_boxes_by_given_contours(polygons_of_marginals)
if not self.curved_line:
if self.light_version: if self.light_version:
if self.textline_light: txt_con_org = get_textregion_contours_in_org_image_light(contours_only_text_parent, self.image, slope_first)
slopes, all_found_textline_polygons, boxes_text, txt_con_org, contours_only_text_parent, all_box_coord, index_by_text_par_con = self.get_slopes_and_deskew_new_light(txt_con_org, contours_only_text_parent, textline_mask_tot_ea_org, image_page_rotated, boxes_text, slope_deskew)
slopes_marginals, all_found_textline_polygons_marginals, boxes_marginals, _, polygons_of_marginals, all_box_coord_marginals, _ = self.get_slopes_and_deskew_new_light(polygons_of_marginals, polygons_of_marginals, textline_mask_tot_ea_org, image_page_rotated, boxes_marginals, slope_deskew)
else:
slopes, all_found_textline_polygons, boxes_text, txt_con_org, contours_only_text_parent, all_box_coord, index_by_text_par_con = self.get_slopes_and_deskew_new_light(txt_con_org, contours_only_text_parent, textline_mask_tot_ea, image_page_rotated, boxes_text, slope_deskew)
slopes_marginals, all_found_textline_polygons_marginals, boxes_marginals, _, polygons_of_marginals, all_box_coord_marginals, _ = self.get_slopes_and_deskew_new_light(polygons_of_marginals, polygons_of_marginals, textline_mask_tot_ea, image_page_rotated, boxes_marginals, slope_deskew)
else: else:
slopes, all_found_textline_polygons, boxes_text, txt_con_org, contours_only_text_parent, all_box_coord, index_by_text_par_con = self.get_slopes_and_deskew_new(txt_con_org, contours_only_text_parent, textline_mask_tot_ea, image_page_rotated, boxes_text, slope_deskew) txt_con_org = get_textregion_contours_in_org_image(contours_only_text_parent, self.image, slope_first)
slopes_marginals, all_found_textline_polygons_marginals, boxes_marginals, _, polygons_of_marginals, all_box_coord_marginals, _ = self.get_slopes_and_deskew_new(polygons_of_marginals, polygons_of_marginals, textline_mask_tot_ea, image_page_rotated, boxes_marginals, slope_deskew) boxes_text, _ = get_text_region_boxes_by_given_contours(contours_only_text_parent)
boxes_marginals, _ = get_text_region_boxes_by_given_contours(polygons_of_marginals)
else: if not self.curved_line:
scale_param = 1
all_found_textline_polygons, boxes_text, txt_con_org, contours_only_text_parent, all_box_coord, index_by_text_par_con, slopes = self.get_slopes_and_deskew_new_curved(txt_con_org, contours_only_text_parent, cv2.erode(textline_mask_tot_ea, kernel=KERNEL, iterations=1), image_page_rotated, boxes_text, text_only, num_col_classifier, scale_param, slope_deskew)
all_found_textline_polygons = small_textlines_to_parent_adherence2(all_found_textline_polygons, textline_mask_tot_ea, num_col_classifier)
all_found_textline_polygons_marginals, boxes_marginals, _, polygons_of_marginals, all_box_coord_marginals, _, slopes_marginals = self.get_slopes_and_deskew_new_curved(polygons_of_marginals, polygons_of_marginals, cv2.erode(textline_mask_tot_ea, kernel=KERNEL, iterations=1), image_page_rotated, boxes_marginals, text_only, num_col_classifier, scale_param, slope_deskew)
all_found_textline_polygons_marginals = small_textlines_to_parent_adherence2(all_found_textline_polygons_marginals, textline_mask_tot_ea, num_col_classifier)
if self.full_layout:
if np.abs(slope_deskew) >= SLOPE_THRESHOLD:
contours_only_text_parent_d_ordered = list(np.array(contours_only_text_parent_d_ordered, dtype=object)[index_by_text_par_con])
if self.light_version: if self.light_version:
text_regions_p, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, slopes, slopes_h, contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered = check_any_text_region_in_model_one_is_main_or_header_light(text_regions_p, regions_fully, contours_only_text_parent, all_box_coord, all_found_textline_polygons, slopes, contours_only_text_parent_d_ordered) if self.textline_light:
slopes, all_found_textline_polygons, boxes_text, txt_con_org, contours_only_text_parent, all_box_coord, index_by_text_par_con = self.get_slopes_and_deskew_new_light(txt_con_org, contours_only_text_parent, textline_mask_tot_ea_org, image_page_rotated, boxes_text, slope_deskew)
slopes_marginals, all_found_textline_polygons_marginals, boxes_marginals, _, polygons_of_marginals, all_box_coord_marginals, _ = self.get_slopes_and_deskew_new_light(polygons_of_marginals, polygons_of_marginals, textline_mask_tot_ea_org, image_page_rotated, boxes_marginals, slope_deskew)
else:
slopes, all_found_textline_polygons, boxes_text, txt_con_org, contours_only_text_parent, all_box_coord, index_by_text_par_con = self.get_slopes_and_deskew_new_light(txt_con_org, contours_only_text_parent, textline_mask_tot_ea, image_page_rotated, boxes_text, slope_deskew)
slopes_marginals, all_found_textline_polygons_marginals, boxes_marginals, _, polygons_of_marginals, all_box_coord_marginals, _ = self.get_slopes_and_deskew_new_light(polygons_of_marginals, polygons_of_marginals, textline_mask_tot_ea, image_page_rotated, boxes_marginals, slope_deskew)
else: else:
text_regions_p, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, slopes, slopes_h, contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered = check_any_text_region_in_model_one_is_main_or_header(text_regions_p, regions_fully, contours_only_text_parent, all_box_coord, all_found_textline_polygons, slopes, contours_only_text_parent_d_ordered) slopes, all_found_textline_polygons, boxes_text, txt_con_org, contours_only_text_parent, all_box_coord, index_by_text_par_con = self.get_slopes_and_deskew_new(txt_con_org, contours_only_text_parent, textline_mask_tot_ea, image_page_rotated, boxes_text, slope_deskew)
slopes_marginals, all_found_textline_polygons_marginals, boxes_marginals, _, polygons_of_marginals, all_box_coord_marginals, _ = self.get_slopes_and_deskew_new(polygons_of_marginals, polygons_of_marginals, textline_mask_tot_ea, image_page_rotated, boxes_marginals, slope_deskew)
else: else:
#takes long timee
contours_only_text_parent_d_ordered = None
if self.light_version:
text_regions_p, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, slopes, slopes_h, contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered = check_any_text_region_in_model_one_is_main_or_header_light(text_regions_p, regions_fully, contours_only_text_parent, all_box_coord, all_found_textline_polygons, slopes, contours_only_text_parent_d_ordered)
else:
text_regions_p, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, slopes, slopes_h, contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered = check_any_text_region_in_model_one_is_main_or_header(text_regions_p, regions_fully, contours_only_text_parent, all_box_coord, all_found_textline_polygons, slopes, contours_only_text_parent_d_ordered)
scale_param = 1
all_found_textline_polygons, boxes_text, txt_con_org, contours_only_text_parent, all_box_coord, index_by_text_par_con, slopes = self.get_slopes_and_deskew_new_curved(txt_con_org, contours_only_text_parent, cv2.erode(textline_mask_tot_ea, kernel=KERNEL, iterations=1), image_page_rotated, boxes_text, text_only, num_col_classifier, scale_param, slope_deskew)
all_found_textline_polygons = small_textlines_to_parent_adherence2(all_found_textline_polygons, textline_mask_tot_ea, num_col_classifier)
all_found_textline_polygons_marginals, boxes_marginals, _, polygons_of_marginals, all_box_coord_marginals, _, slopes_marginals = self.get_slopes_and_deskew_new_curved(polygons_of_marginals, polygons_of_marginals, cv2.erode(textline_mask_tot_ea, kernel=KERNEL, iterations=1), image_page_rotated, boxes_marginals, text_only, num_col_classifier, scale_param, slope_deskew)
all_found_textline_polygons_marginals = small_textlines_to_parent_adherence2(all_found_textline_polygons_marginals, textline_mask_tot_ea, num_col_classifier)
if self.full_layout:
if np.abs(slope_deskew) >= SLOPE_THRESHOLD:
contours_only_text_parent_d_ordered = list(np.array(contours_only_text_parent_d_ordered, dtype=object)[index_by_text_par_con])
if self.light_version:
text_regions_p, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, slopes, slopes_h, contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered = check_any_text_region_in_model_one_is_main_or_header_light(text_regions_p, regions_fully, contours_only_text_parent, all_box_coord, all_found_textline_polygons, slopes, contours_only_text_parent_d_ordered)
else:
text_regions_p, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, slopes, slopes_h, contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered = check_any_text_region_in_model_one_is_main_or_header(text_regions_p, regions_fully, contours_only_text_parent, all_box_coord, all_found_textline_polygons, slopes, contours_only_text_parent_d_ordered)
else:
#takes long timee
contours_only_text_parent_d_ordered = None
if self.light_version:
text_regions_p, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, slopes, slopes_h, contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered = check_any_text_region_in_model_one_is_main_or_header_light(text_regions_p, regions_fully, contours_only_text_parent, all_box_coord, all_found_textline_polygons, slopes, contours_only_text_parent_d_ordered)
else:
text_regions_p, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, slopes, slopes_h, contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered = check_any_text_region_in_model_one_is_main_or_header(text_regions_p, regions_fully, contours_only_text_parent, all_box_coord, all_found_textline_polygons, slopes, contours_only_text_parent_d_ordered)
if self.plotter:
self.plotter.save_plot_of_layout(text_regions_p, image_page)
self.plotter.save_plot_of_layout_all(text_regions_p, image_page)
pixel_img = 4
polygons_of_drop_capitals = return_contours_of_interested_region_by_min_size(text_regions_p, pixel_img)
all_found_textline_polygons = adhere_drop_capital_region_into_corresponding_textline(text_regions_p, polygons_of_drop_capitals, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, kernel=KERNEL, curved_line=self.curved_line)
pixel_lines = 6
if not self.headers_off:
if np.abs(slope_deskew) < SLOPE_THRESHOLD:
num_col, _, matrix_of_lines_ch, splitter_y_new, _ = find_number_of_columns_in_document(np.repeat(text_regions_p[:, :, np.newaxis], 3, axis=2), num_col_classifier, self.tables, pixel_lines, contours_only_text_parent_h)
else:
_, _, matrix_of_lines_ch_d, splitter_y_new_d, _ = find_number_of_columns_in_document(np.repeat(text_regions_p_1_n[:, :, np.newaxis], 3, axis=2), num_col_classifier, self.tables, pixel_lines, contours_only_text_parent_h_d_ordered)
elif self.headers_off:
if np.abs(slope_deskew) < SLOPE_THRESHOLD:
num_col, _, matrix_of_lines_ch, splitter_y_new, _ = find_number_of_columns_in_document(np.repeat(text_regions_p[:, :, np.newaxis], 3, axis=2), num_col_classifier, self.tables, pixel_lines)
else:
_, _, matrix_of_lines_ch_d, splitter_y_new_d, _ = find_number_of_columns_in_document(np.repeat(text_regions_p_1_n[:, :, np.newaxis], 3, axis=2), num_col_classifier, self.tables, pixel_lines)
if num_col_classifier >= 3:
if np.abs(slope_deskew) < SLOPE_THRESHOLD:
regions_without_separators = regions_without_separators.astype(np.uint8)
regions_without_separators = cv2.erode(regions_without_separators[:, :], KERNEL, iterations=6)
else:
regions_without_separators_d = regions_without_separators_d.astype(np.uint8)
regions_without_separators_d = cv2.erode(regions_without_separators_d[:, :], KERNEL, iterations=6)
if np.abs(slope_deskew) < SLOPE_THRESHOLD:
boxes, peaks_neg_tot_tables = return_boxes_of_images_by_order_of_reading_new(splitter_y_new, regions_without_separators, matrix_of_lines_ch, num_col_classifier, erosion_hurts, self.tables, self.right2left)
else:
boxes_d, peaks_neg_tot_tables_d = return_boxes_of_images_by_order_of_reading_new(splitter_y_new_d, regions_without_separators_d, matrix_of_lines_ch_d, num_col_classifier, erosion_hurts, self.tables, self.right2left)
#print(boxes_d,'boxes_d')
#img_once = np.zeros((textline_mask_tot_d.shape[0],textline_mask_tot_d.shape[1]))
#for box_i in boxes_d:
#img_once[int(box_i[2]):int(box_i[3]),int(box_i[0]):int(box_i[1]) ] =1
#plt.imshow(img_once)
#plt.show()
#print(np.unique(img_once),'img_once')
if self.plotter: if self.plotter:
self.plotter.save_plot_of_layout(text_regions_p, image_page) self.plotter.write_images_into_directory(polygons_of_images, image_page)
self.plotter.save_plot_of_layout_all(text_regions_p, image_page) t_order = time.time()
if self.full_layout:
pixel_img = 4
polygons_of_drop_capitals = return_contours_of_interested_region_by_min_size(text_regions_p, pixel_img)
all_found_textline_polygons = adhere_drop_capital_region_into_corresponding_textline(text_regions_p, polygons_of_drop_capitals, contours_only_text_parent, contours_only_text_parent_h, all_box_coord, all_box_coord_h, all_found_textline_polygons, all_found_textline_polygons_h, kernel=KERNEL, curved_line=self.curved_line)
pixel_lines = 6
if not self.headers_off:
if np.abs(slope_deskew) < SLOPE_THRESHOLD: if np.abs(slope_deskew) < SLOPE_THRESHOLD:
num_col, _, matrix_of_lines_ch, splitter_y_new, _ = find_number_of_columns_in_document(np.repeat(text_regions_p[:, :, np.newaxis], 3, axis=2), num_col_classifier, self.tables, pixel_lines, contours_only_text_parent_h) order_text_new, id_of_texts_tot = self.do_order_of_regions(contours_only_text_parent, contours_only_text_parent_h, boxes, textline_mask_tot)
else: else:
_, _, matrix_of_lines_ch_d, splitter_y_new_d, _ = find_number_of_columns_in_document(np.repeat(text_regions_p_1_n[:, :, np.newaxis], 3, axis=2), num_col_classifier, self.tables, pixel_lines, contours_only_text_parent_h_d_ordered) order_text_new, id_of_texts_tot = self.do_order_of_regions(contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered, boxes_d, textline_mask_tot_d)
elif self.headers_off:
pcgts = self.writer.build_pagexml_full_layout(contours_only_text_parent, contours_only_text_parent_h, page_coord, order_text_new, id_of_texts_tot, all_found_textline_polygons, all_found_textline_polygons_h, all_box_coord, all_box_coord_h, polygons_of_images, contours_tables, polygons_of_drop_capitals, polygons_of_marginals, all_found_textline_polygons_marginals, all_box_coord_marginals, slopes, slopes_h, slopes_marginals, cont_page, polygons_lines_xml)
self.logger.info("Job done in %.1fs", time.time() - t0)
if not self.dir_in:
return pcgts
else:
contours_only_text_parent_h = None
if np.abs(slope_deskew) < SLOPE_THRESHOLD: if np.abs(slope_deskew) < SLOPE_THRESHOLD:
num_col, _, matrix_of_lines_ch, splitter_y_new, _ = find_number_of_columns_in_document(np.repeat(text_regions_p[:, :, np.newaxis], 3, axis=2), num_col_classifier, self.tables, pixel_lines) order_text_new, id_of_texts_tot = self.do_order_of_regions(contours_only_text_parent, contours_only_text_parent_h, boxes, textline_mask_tot)
else: else:
_, _, matrix_of_lines_ch_d, splitter_y_new_d, _ = find_number_of_columns_in_document(np.repeat(text_regions_p_1_n[:, :, np.newaxis], 3, axis=2), num_col_classifier, self.tables, pixel_lines) contours_only_text_parent_d_ordered = list(np.array(contours_only_text_parent_d_ordered, dtype=object)[index_by_text_par_con])
order_text_new, id_of_texts_tot = self.do_order_of_regions(contours_only_text_parent_d_ordered, contours_only_text_parent_h, boxes_d, textline_mask_tot_d)
if num_col_classifier >= 3: pcgts = self.writer.build_pagexml_no_full_layout(txt_con_org, page_coord, order_text_new, id_of_texts_tot, all_found_textline_polygons, all_box_coord, polygons_of_images, polygons_of_marginals, all_found_textline_polygons_marginals, all_box_coord_marginals, slopes, slopes_marginals, cont_page, polygons_lines_xml, contours_tables)
if np.abs(slope_deskew) < SLOPE_THRESHOLD: self.logger.info("Job done in %.1fs", time.time() - t0)
regions_without_separators = regions_without_separators.astype(np.uint8) if not self.dir_in:
regions_without_separators = cv2.erode(regions_without_separators[:, :], KERNEL, iterations=6) return pcgts
else:
regions_without_separators_d = regions_without_separators_d.astype(np.uint8)
regions_without_separators_d = cv2.erode(regions_without_separators_d[:, :], KERNEL, iterations=6)
if np.abs(slope_deskew) < SLOPE_THRESHOLD:
boxes, peaks_neg_tot_tables = return_boxes_of_images_by_order_of_reading_new(splitter_y_new, regions_without_separators, matrix_of_lines_ch, num_col_classifier, erosion_hurts, self.tables, self.right2left)
else:
boxes_d, peaks_neg_tot_tables_d = return_boxes_of_images_by_order_of_reading_new(splitter_y_new_d, regions_without_separators_d, matrix_of_lines_ch_d, num_col_classifier, erosion_hurts, self.tables, self.right2left)
#print(boxes_d,'boxes_d')
#img_once = np.zeros((textline_mask_tot_d.shape[0],textline_mask_tot_d.shape[1]))
#for box_i in boxes_d:
#img_once[int(box_i[2]):int(box_i[3]),int(box_i[0]):int(box_i[1]) ] =1
#plt.imshow(img_once)
#plt.show()
#print(np.unique(img_once),'img_once')
if self.plotter:
self.plotter.write_images_into_directory(polygons_of_images, image_page)
t_order = time.time()
if self.full_layout:
if np.abs(slope_deskew) < SLOPE_THRESHOLD:
order_text_new, id_of_texts_tot = self.do_order_of_regions(contours_only_text_parent, contours_only_text_parent_h, boxes, textline_mask_tot)
else:
order_text_new, id_of_texts_tot = self.do_order_of_regions(contours_only_text_parent_d_ordered, contours_only_text_parent_h_d_ordered, boxes_d, textline_mask_tot_d)
pcgts = self.writer.build_pagexml_full_layout(contours_only_text_parent, contours_only_text_parent_h, page_coord, order_text_new, id_of_texts_tot, all_found_textline_polygons, all_found_textline_polygons_h, all_box_coord, all_box_coord_h, polygons_of_images, contours_tables, polygons_of_drop_capitals, polygons_of_marginals, all_found_textline_polygons_marginals, all_box_coord_marginals, slopes, slopes_h, slopes_marginals, cont_page, polygons_lines_xml)
self.logger.info("Job done in %.1fs", time.time() - t0)
if not self.dir_in:
return pcgts
else:
contours_only_text_parent_h = None
if np.abs(slope_deskew) < SLOPE_THRESHOLD:
order_text_new, id_of_texts_tot = self.do_order_of_regions(contours_only_text_parent, contours_only_text_parent_h, boxes, textline_mask_tot)
else:
contours_only_text_parent_d_ordered = list(np.array(contours_only_text_parent_d_ordered, dtype=object)[index_by_text_par_con])
order_text_new, id_of_texts_tot = self.do_order_of_regions(contours_only_text_parent_d_ordered, contours_only_text_parent_h, boxes_d, textline_mask_tot_d)
pcgts = self.writer.build_pagexml_no_full_layout(txt_con_org, page_coord, order_text_new, id_of_texts_tot, all_found_textline_polygons, all_box_coord, polygons_of_images, polygons_of_marginals, all_found_textline_polygons_marginals, all_box_coord_marginals, slopes, slopes_marginals, cont_page, polygons_lines_xml, contours_tables)
self.logger.info("Job done in %.1fs", time.time() - t0)
if not self.dir_in:
return pcgts
if self.dir_in:
self.writer.write_pagexml(pcgts)
#self.logger.info("Job done in %.1fs", time.time() - t0)
if self.dir_in: if self.dir_in:
self.writer.write_pagexml(pcgts) self.logger.info("All jobs done in %.1fs", time.time() - t0_tot)
#self.logger.info("Job done in %.1fs", time.time() - t0)
if self.dir_in:
self.logger.info("All jobs done in %.1fs", time.time() - t0_tot)

6
src/eynollah/ocrd-tool.json
View file

@ -77,10 +77,10 @@
}, },
"resources": [ "resources": [
{ {
"description": "models for eynollah (TensorFlow format)", "description": "models for eynollah (TensorFlow SavedModel format)",
"url": "https://github.com/qurator-spk/eynollah/releases/download/v0.3.0/models_eynollah.tar.gz", "url": "https://github.com/qurator-spk/eynollah/releases/download/v0.3.1/models_eynollah.tar.gz",
"name": "default", "name": "default",
"size": 1761991295, "size": 1894627041,
"type": "archive", "type": "archive",
"path_in_archive": "models_eynollah" "path_in_archive": "models_eynollah"
} }

16
src/eynollah/writer.py
View file

@ -172,10 +172,18 @@ class EynollahXmlWriter():
page.add_ImageRegion(img_region) page.add_ImageRegion(img_region)
points_co = '' points_co = ''
for lmm in range(len(found_polygons_text_region_img[mm])): for lmm in range(len(found_polygons_text_region_img[mm])):
points_co += str(int((found_polygons_text_region_img[mm][lmm,0,0] + page_coord[2]) / self.scale_x)) try:
points_co += ',' points_co += str(int((found_polygons_text_region_img[mm][lmm,0,0] + page_coord[2]) / self.scale_x))
points_co += str(int((found_polygons_text_region_img[mm][lmm,0,1] + page_coord[0]) / self.scale_y)) points_co += ','
points_co += ' ' points_co += str(int((found_polygons_text_region_img[mm][lmm,0,1] + page_coord[0]) / self.scale_y))
points_co += ' '
except:
points_co += str(int((found_polygons_text_region_img[mm][lmm][0] + page_coord[2])/ self.scale_x ))
points_co += ','
points_co += str(int((found_polygons_text_region_img[mm][lmm][1] + page_coord[0])/ self.scale_y ))
points_co += ' '
img_region.get_Coords().set_points(points_co[:-1]) img_region.get_Coords().set_points(points_co[:-1])
for mm in range(len(polygons_lines_to_be_written_in_xml)): for mm in range(len(polygons_lines_to_be_written_in_xml)):