Marmelator/cv-analysis-service
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 37 Wiki Activity

ready for integration with pyinfra

Browse Source
This commit is contained in:
Isaac Riley 2022-06-13 12:59:00 +02:00
parent 01803d452a
commit c62ab08b98
38 changed files with 4160 additions and 359 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
config.yaml
View File

@ -23,5 +23,5 @@ deskew:
test_dummy: test_dummy
visual_logging:
level: $LOGGING_LEVEL_ROOT|DEBUG
level: $LOGGING_LEVEL_ROOT|INFO # NOTHNG > INFO > DEBUG > ALL
output_folder: /tmp/debug/

58
cv_analysis/fig_detection_with_layout.py
View File

@ -1,58 +0,0 @@
from cv_analysis.layout_parsing import annotate_layout_in_pdf
from cv_analysis.figure_detection import detect_figures
from cv_analysis.table_parsing import tables_in_image, parse_table
from cv_analysis.utils.draw import draw_rectangles
from cv_analysis.utils.display import show_mpl
from cv_analysis.utils.visual_logging import vizlogger
#from PIL import Image
def cut_out_content_structures(layout_rects, page):
large_rects = []
small_rects = []
for x, y, w, h in layout_rects:
rect = (x, y, w, h)
if w * h >= 75000:
cropped_page = page[y:(y + h), x:(x + w)]
large_rects.append([rect, cropped_page])
else:
cropped_page = page[y:(y + h), x:(x + w)]
small_rects.append([rect, cropped_page])
return large_rects, small_rects
def parse_content_structures(page, large_rects, small_rects):
for coordinates, cropped_image in large_rects:
figure_rects = detect_figures(cropped_image)
if len(figure_rects) == 0: # text
page = draw_rectangles(page, [coordinates], color=(0, 255, 0), annotate=True)
elif len(parse_table(cropped_image)) > 0:
#elif tables_in_image(cropped_image)[0]: # table
stats = parse_table(cropped_image)
cropped_image = draw_rectangles(cropped_image, stats, color=(255, 0, 0), annotate=True)
x,y,w,h = coordinates
page[y:y+h, x:x+w] = cropped_image
else: # figure
page = draw_rectangles(page, [coordinates], color=(0, 0, 255), annotate=True)
# for coordinates, cropped_image in small_rects:
# figure_rects = detect_figures(cropped_image)
# if len(figure_rects) == 0 and len(list(find_primary_text_regions(cropped_image))) > 0:
# page = draw_rectangles(page, [coordinates], color=(0, 255, 0), annotate=True)
# else:
# page = draw_rectangles(page, [coordinates], color=(0, 255, 255), annotate=True)
return page
def detect_figures_with_layout_parsing(pdf_path, page_index=1, show=False):
layout_rects, page = annotate_layout_in_pdf(pdf_path, page_index, return_rects=True)
big_structures, small_structures = cut_out_content_structures(layout_rects, page)
page = parse_content_structures(page, big_structures, small_structures)
vizlogger.debug(page, "figures03_final.png")
if show:
show_mpl(page)
else:
return page

70
cv_analysis/figure_detection.py
View File

@ -1,7 +1,7 @@
import cv2
import numpy as np
from pdf2image import pdf2image
import pandas as pd
#import pandas as pd
from PIL import Image
import timeit
from os import path
@ -11,6 +11,7 @@ from cv_analysis.utils.display import show_mpl
from cv_analysis.utils.draw import draw_rectangles
from cv_analysis.utils.post_processing import remove_included
from cv_analysis.utils.filters import is_large_enough, has_acceptable_format
from cv_analysis.utils.structures import Rectangle
from cv_analysis.utils.text import remove_primary_text_regions
from cv_analysis.utils.visual_logging import vizlogger
@ -32,69 +33,4 @@ def detect_figures(image: np.array):
rects = map(cv2.boundingRect, cnts)
rects = remove_included(rects)
return list(rects)
def detect_figures_in_pdf(pdf_path, page_index=1, show=False):
page = pdf2image.convert_from_path(pdf_path, dpi=300, first_page=page_index + 1, last_page=page_index + 1)[0]
page = np.array(page)
redaction_contours = detect_figures(page)
page = draw_rectangles(page, redaction_contours)
vizlogger.debug(page, "figures03_final.png")
if show:
show_mpl(page)
return page
def detect_figures_in_test_files():
def save_as_pdf(pages):
p1, p = pages[0], pages[1:]
out_pdf_path = "/home/lillian/ocr_docs/output_files/fig_detection_pdf.pdf"
p1.save(
out_pdf_path, "PDF", resolution=150.0, save_all=True, append_images=p
)
path = "/home/lillian/ocr_docs/"
ex_pages = pd.read_csv(path+"/metadata/metadata2.csv")
pages_detected = []
t0 = timeit.default_timer()
for name, page_nr in zip(ex_pages.pdf_name, ex_pages.page):
page = pdf2image.convert_from_path(path + "/original/" + name, dpi=300, first_page=page_nr, last_page=page_nr)[0]
page = np.array(page)
redaction_contours = detect_figures(page)
page = draw_rectangles(page, redaction_contours)
pages_detected.append(Image.fromarray(page))
print(timeit.default_timer()-t0)
save_as_pdf(pages_detected)
def detect_figures_in_png(pdf_path, show=False):
page = Image.open(pdf_path)
page = np.array(page)
redaction_contours = detect_figures(page)
page = draw_rectangles(page, redaction_contours)
vizlogger.debug(page, "figures03_final.png")
if show:
show_mpl(page)
return page
def detect_figures_in_test_files_png():
file_name = pd.read_csv(METADATA_TESTFILES)
pages = []
t0 = timeit.default_timer()
for name in file_name.image_name:
page = detect_figures_in_png(path.join(PNG_FOR_TESTING, name+".png"))
pages.append(Image.fromarray(page))
t1 = timeit.default_timer()
print(t1-t0)
p1, p = pages[0], pages[1:]
out_pdf_path = path.join(PNG_FIGURES_DETECTED, "fig_detectes.pdf")
p1.save(
out_pdf_path, "PDF", resolution=300.0, save_all=True, append_images=p
)
return list(map(Rectangle.from_xywh, rects))

101
cv_analysis/layout_parsing.py
View File

@ -4,10 +4,11 @@ from operator import __and__
import cv2
import numpy as np
from pdf2image import pdf2image
#from pdf2image import pdf2image
from cv_analysis.utils.display import show_mpl
from cv_analysis.utils.draw import draw_rectangles
#from cv_analysis.utils.display import show_mpl
#from cv_analysis.utils.draw import draw_rectangles
from cv_analysis.utils.structures import Rectangle
from cv_analysis.utils.post_processing import remove_overlapping, remove_included, has_no_parent
from cv_analysis.utils.visual_logging import vizlogger
@ -36,17 +37,17 @@ def parse_layout(image: np.array):
if len(image_.shape) > 2:
image_ = cv2.cvtColor(image_, cv2.COLOR_BGR2GRAY)
#vizlogger.debug(image_, "layout01_start.png")
vizlogger.debug(image_, "layout01_start.png")
image_ = cv2.GaussianBlur(image_, (7, 7), 0)
#vizlogger.debug(image_, "layout02_blur.png")
vizlogger.debug(image_, "layout02_blur.png")
thresh = cv2.threshold(image_, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]
vizlogger.debug(image_, "layout03_theshold.png")
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 5))
#vizlogger.debug(kernel, "layout04_kernel.png")
vizlogger.debug(kernel, "layout04_kernel.png")
dilate = cv2.dilate(thresh, kernel, iterations=4)
#vizlogger.debug(dilate, "layout05_dilate.png")
vizlogger.debug(dilate, "layout05_dilate.png")
rects = list(find_segments(dilate))
@ -55,16 +56,16 @@ def parse_layout(image: np.array):
x, y, w, h = rect
cv2.rectangle(image, (x, y), (x + w, y + h), (0, 0, 0), -1)
cv2.rectangle(image, (x, y), (x + w, y + h), (255, 255, 255), 7)
#vizlogger.debug(image, "layout06_rectangles.png")
vizlogger.debug(image, "layout06_rectangles.png")
_, image = cv2.threshold(image, 254, 255, cv2.THRESH_BINARY)
#vizlogger.debug(image, "layout07_threshold.png")
vizlogger.debug(image, "layout07_threshold.png")
image = ~image
#vizlogger.debug(image, "layout08_inverse.png")
vizlogger.debug(image, "layout08_inverse.png")
if len(image.shape) > 2:
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
#vizlogger.debug(image, "layout09_convertcolor.png")
vizlogger.debug(image, "layout09_convertcolor.png")
rects = find_segments(image)
# <- End of meta detection
@ -72,73 +73,23 @@ def parse_layout(image: np.array):
rects = remove_included(rects)
rects = remove_overlapping(rects)
return list(rects)
return list(map(Rectangle.from_xywh, rects))
def annotate_layout_in_pdf(pdf_path, page_index=1, return_rects=False, show=False):
# def annotate_layout_in_pdf(page, return_rects=False, show=False):
page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
page = np.array(page)
# #page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
# #page = np.array(page)
rects = parse_layout(page)
# rects = parse_layout(page)
if return_rects:
return rects, page
elif show:
page = draw_rectangles(page, rects)
vizlogger.debug(page, "layout10_output.png")
show_mpl(page)
else:
page = draw_rectangles(page, rects)
return page
# if return_rects:
# return rects, page
# elif show:
# page = draw_rectangles(page, rects)
# vizlogger.debug(page, "layout10_output.png")
# show_mpl(page)
# else:
# page = draw_rectangles(page, rects)
# return page
"""
def find_layout_boxes(image: np.array):
if len(image.shape) > 2:
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
image = cv2.GaussianBlur(image, (5, 5), 1)
image = cv2.threshold(image, 253, 255, cv2.THRESH_BINARY)[1]
img_bin = ~image
line_min_width = 10
kernel_h = np.ones((10, line_min_width), np.uint8)
kernel_v = np.ones((line_min_width, 10), np.uint8)
img_bin_h = cv2.morphologyEx(img_bin, cv2.MORPH_OPEN, kernel_h)
img_bin_v = cv2.morphologyEx(img_bin, cv2.MORPH_OPEN, kernel_v)
img_bin_final = img_bin_h | img_bin_v
contours = cv2.findContours(img_bin_final, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
contours = imutils.grab_contours(contours)
for c in contours:
peri = cv2.arcLength(c, True)
approx = cv2.approxPolyDP(c, 0.04 * peri, True)
yield cv2.boundingRect(approx)
def annotate_layout_boxes(image, rects):
for rect in rects:
(x, y, w, h) = rect
cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
return image
def annotate_layout_in_pdf(pdf_path, page_index=1):
page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
page = np.array(page)
layout_boxes = find_layout_boxes(page)
page = annotate_layout_boxes(page, layout_boxes)
fig, ax = plt.subplots(1, 1)
fig.set_size_inches(20, 20)
ax.imshow(page)
plt.show()
"""

16
cv_analysis/redaction_detection.py
View File

@ -40,14 +40,14 @@ def find_redactions(image: np.array, min_normalized_area=200000):
return []
def annotate_redactions_in_pdf(pdf_path, page_index=1, show=False):
# def annotate_redactions_in_pdf(page, show=False):
page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
page = np.array(page)
# #page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
# #page = np.array(page)
redaction_contours = find_redactions(page)
page = draw_contours(page, redaction_contours)
vizlogger.debug(page, "redactions05_output.png")
# redaction_contours = find_redactions(page)
# page = draw_contours(page, redaction_contours)
# vizlogger.debug(page, "redactions05_output.png")
if show:
show_mpl(page)
# if show:
# show_mpl(page)

64
cv_analysis/table_parsing.py
View File

@ -2,19 +2,16 @@
from functools import partial
from itertools import chain, starmap
from operator import attrgetter
import cv2
import numpy as np
from pdf2image import pdf2image
from cv_analysis.utils.display import show_mpl
from cv_analysis.utils.draw import draw_rectangles
#from pdf2image import pdf2image
#from cv_analysis.utils.display import show_mpl
#from cv_analysis.utils.draw import draw_rectangles
from cv_analysis.utils.post_processing import xywh_to_vecs, xywh_to_vec_rect, adjacent1d
from cv_analysis.utils.deskew import deskew_histbased, deskew
from cv_analysis.utils.filters import is_large_enough
#from cv_analysis.utils.deskew import deskew_histbased, deskew
#from cv_analysis.utils.filters import is_large_enough
from cv_analysis.utils.structures import Rectangle
from cv_analysis.utils.visual_logging import vizlogger
from cv_analysis.layout_parsing import parse_layout
@ -138,7 +135,7 @@ def find_table_layout_boxes(image: np.array):
for box in layout_boxes:
(x, y, w, h) = box
if w * h >= 100000:
table_boxes.append(box)
table_boxes.append(Rectangle.from_xywh(box))
return table_boxes
@ -148,7 +145,7 @@ def preprocess(image: np.array):
return ~image
def parse_table(image: np.array, show=False):
def parse_tables(image: np.array, show=False):
"""Runs the full table parsing process.
Args:
@ -164,42 +161,33 @@ def parse_table(image: np.array, show=False):
image = preprocess(image)
table_layout_boxes = find_table_layout_boxes(image)
#table_layout_boxes = find_table_layout_boxes(image)
image = isolate_vertical_and_horizontal_components(image)
# image = add_external_contours(image, image)
# vizlogger.debug(image, "external_contours_added.png")
vizlogger.debug(image, "external_contours_added.png")
_, _, stats, _ = cv2.connectedComponentsWithStats(~image, connectivity=8, ltype=cv2.CV_32S)
stats = np.vstack(list(filter(is_large_enough, stats)))
rects = stats[:, :-1][2:]
return list(map(list, rects))
#print(rects)
return list(map(Rectangle.from_xywh, rects))
def annotate_tables_in_pdf(pdf_path, page_index=0, deskew=False, show=False):
""" """
page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
page = np.array(page)
if show:
show_mpl(page)
if deskew:
page, _ = deskew_histbased(page)
stats = parse_table(page)
page = draw_rectangles(page, stats, annotate=True)
vizlogger.debug(page, "tables15_final_output.png")
if show:
show_mpl(page)
def tables_in_image(cropped_image):
table_rects = parse_table(cropped_image)
if len(table_rects) > 0:
return True, table_rects
else:
return False, None
# def annotate_tables_in_pdf(page, page_index=0, deskew=False, show=False):
# """ """
# #page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
# #page = np.array(page)
# if show:
# show_mpl(page)
# if deskew:
# page, _ = deskew_histbased(page)
# stats = parse_tables(page)
# page = draw_rectangles(page, stats, annotate=True)
# vizlogger.debug(page, "tables15_final_output.png")
# if show:
# show_mpl(page)

54
cv_analysis/test/test_data/ref.json Normal file
View File

@ -0,0 +1,54 @@
{
"images": [
{
"name": "test1.png",
"source_document": "Amended Residue analytical method for the determ.pdf",
"page": 7
},
{
"name": "test2.png",
"source_document": "Amended Residue analytical method for the determ.pdf",
"page": 39
},
{
"name": "test3.png",
"source_document": "VV-857853.pdf",
"page": 8
},
{
"name": "test4.png",
"source_document": "Sulphur_RAR_09_Volume_3CA_B-7_2021-04-09.pdf",
"page": 25
},
{
"name": "test5.png",
"source_document": "Sulphur_RAR_09_Volume_3CA_B-7_2021-04-09.pdf",
"page": 35
},
{
"name": "test6.png",
"source_document": "VV-128279.pdf",
"page": 49
},
{
"name": "test7.png",
"source_document": "VV-376573.pdf",
"page": 86
},
{
"name": "test8.png",
"source_document": "VV-377325.pdf",
"page": 218
},
{
"name": "test9.png",
"source_document": "VV-857853.pdf",
"page": 10
},
{
"name": "test10.png",
"source_document": "VV-334103.pdf",
"page": 28
}
]
}

419
cv_analysis/test/test_data/table.json
View File

@ -1,62 +1,359 @@
{
"0": [
[211, 447, 367, 47],
[581, 447, 417, 47],
[1001, 447, 406, 47],
[211, 497, 367, 47],
[580, 497, 418, 47],
[1001, 497, 406, 47],
[211, 547, 367, 47],
[580, 547, 418, 47],
[1001, 547, 406, 47],
[211, 597, 367, 47],
[581, 597, 417, 47],
[1001, 597, 406, 48],
[212, 647, 366, 48],
[581, 647, 417, 48],
[1001, 647, 406, 48],
[581, 697, 417, 47],
[1001, 697, 407, 48],
[212, 698, 366, 47],
[211, 747, 367, 48],
[581, 747, 417, 48],
[1001, 748, 407, 47],
[211, 798, 367, 47],
[581, 798, 417, 47],
[1001, 798, 407, 47],
[212, 848, 366, 47],
[581, 848, 417, 47],
[1001, 848, 407, 48],
[212, 898, 366, 48],
[581, 898, 417, 48],
[1001, 898, 407, 48],
[462, 1195, 368, 48],
[833, 1195, 404, 48],
[462, 1245, 368, 48],
[833, 1245, 404, 47],
[462, 1296, 368, 47],
[833, 1296, 404, 47],
[462, 1346, 368, 47],
[833, 1346, 404, 47],
[462, 1396, 368, 47],
[834, 1396, 403, 47],
[462, 1446, 368, 48],
[833, 1446, 404, 48],
[462, 1496, 368, 48],
[833, 1496, 404, 48],
[462, 1547, 368, 47],
[834, 1547, 403, 47],
[462, 1597, 368, 48],
[834, 1597, 403, 47],
[462, 1647, 368, 48],
[833, 1647, 404, 48],
[462, 1698, 368, 47],
[833, 1698, 404, 47],
[462, 1748, 368, 47],
[834, 1748, 403, 47],
[462, 1798, 368, 47],
[834, 1798, 403, 47],
[462, 1848, 368, 48],
[834, 1848, 403, 48]
]
}
"pages": [
{
"page": 0,
"pageWidth": 2346,
"pageHeight": 1663,
"cells": [
{
"x": 211,
"y": 447,
"width": 367,
"height": 47
},
{
"x": 581,
"y": 447,
"width": 417,
"height": 47
},
{
"x": 1001,
"y": 447,
"width": 406,
"height": 47
},
{
"x": 211,
"y": 497,
"width": 367,
"height": 47
},
{
"x": 580,
"y": 497,
"width": 418,
"height": 47
},
{
"x": 1001,
"y": 497,
"width": 406,
"height": 47
},
{
"x": 211,
"y": 547,
"width": 367,
"height": 47
},
{
"x": 580,
"y": 547,
"width": 418,
"height": 47
},
{
"x": 1001,
"y": 547,
"width": 406,
"height": 47
},
{
"x": 211,
"y": 597,
"width": 367,
"height": 47
},
{
"x": 581,
"y": 597,
"width": 417,
"height": 47
},
{
"x": 1001,
"y": 597,
"width": 406,
"height": 48
},
{
"x": 212,
"y": 647,
"width": 366,
"height": 48
},
{
"x": 581,
"y": 647,
"width": 417,
"height": 48
},
{
"x": 1001,
"y": 647,
"width": 406,
"height": 48
},
{
"x": 581,
"y": 697,
"width": 417,
"height": 47
},
{
"x": 1001,
"y": 697,
"width": 407,
"height": 48
},
{
"x": 212,
"y": 698,
"width": 366,
"height": 47
},
{
"x": 211,
"y": 747,
"width": 367,
"height": 48
},
{
"x": 581,
"y": 747,
"width": 417,
"height": 48
},
{
"x": 1001,
"y": 748,
"width": 407,
"height": 47
},
{
"x": 211,
"y": 798,
"width": 367,
"height": 47
},
{
"x": 581,
"y": 798,
"width": 417,
"height": 47
},
{
"x": 1001,
"y": 798,
"width": 407,
"height": 47
},
{
"x": 212,
"y": 848,
"width": 366,
"height": 47
},
{
"x": 581,
"y": 848,
"width": 417,
"height": 47
},
{
"x": 1001,
"y": 848,
"width": 407,
"height": 48
},
{
"x": 212,
"y": 898,
"width": 366,
"height": 48
},
{
"x": 581,
"y": 898,
"width": 417,
"height": 48
},
{
"x": 1001,
"y": 898,
"width": 407,
"height": 48
},
{
"x": 462,
"y": 1195,
"width": 368,
"height": 48
},
{
"x": 833,
"y": 1195,
"width": 404,
"height": 48
},
{
"x": 462,
"y": 1245,
"width": 368,
"height": 48
},
{
"x": 833,
"y": 1245,
"width": 404,
"height": 47
},
{
"x": 462,
"y": 1296,
"width": 368,
"height": 47
},
{
"x": 833,
"y": 1296,
"width": 404,
"height": 47
},
{
"x": 462,
"y": 1346,
"width": 368,
"height": 47
},
{
"x": 833,
"y": 1346,
"width": 404,
"height": 47
},
{
"x": 462,
"y": 1396,
"width": 368,
"height": 47
},
{
"x": 834,
"y": 1396,
"width": 403,
"height": 47
},
{
"x": 462,
"y": 1446,
"width": 368,
"height": 48
},
{
"x": 833,
"y": 1446,
"width": 404,
"height": 48
},
{
"x": 462,
"y": 1496,
"width": 368,
"height": 48
},
{
"x": 833,
"y": 1496,
"width": 404,
"height": 48
},
{
"x": 462,
"y": 1547,
"width": 368,
"height": 47
},
{
"x": 834,
"y": 1547,
"width": 403,
"height": 47
},
{
"x": 462,
"y": 1597,
"width": 368,
"height": 48
},
{
"x": 834,
"y": 1597,
"width": 403,
"height": 47
},
{
"x": 462,
"y": 1647,
"width": 368,
"height": 48
},
{
"x": 833,
"y": 1647,
"width": 404,
"height": 48
},
{
"x": 462,
"y": 1698,
"width": 368,
"height": 47
},
{
"x": 833,
"y": 1698,
"width": 404,
"height": 47
},
{
"x": 462,
"y": 1748,
"width": 368,
"height": 47
},
{
"x": 834,
"y": 1748,
"width": 403,
"height": 47
},
{
"x": 462,
"y": 1798,
"width": 368,
"height": 47
},
{
"x": 834,
"y": 1798,
"width": 403,
"height": 47
},
{
"x": 462,
"y": 1848,
"width": 368,
"height": 48
},
{
"x": 834,
"y": 1848,
"width": 403,
"height": 48
}
]
}
]
}

191
cv_analysis/test/test_data/test1.json Normal file
View File

@ -0,0 +1,191 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2481,
"pageHeight": 3509,
"cells": [
{
"x": 604,
"y": 400,
"width": 399,
"height": 142
},
{
"x": 1006,
"y": 400,
"width": 49,
"height": 142
},
{
"x": 1058,
"y": 400,
"width": 1215,
"height": 142
},
{
"x": 604,
"y": 545,
"width": 399,
"height": 83
},
{
"x": 1006,
"y": 545,
"width": 49,
"height": 83
},
{
"x": 1058,
"y": 545,
"width": 1215,
"height": 83
},
{
"x": 604,
"y": 631,
"width": 399,
"height": 84
},
{
"x": 1006,
"y": 631,
"width": 49,
"height": 84
},
{
"x": 1058,
"y": 631,
"width": 1215,
"height": 84
},
{
"x": 604,
"y": 718,
"width": 399,
"height": 84
},
{
"x": 1006,
"y": 718,
"width": 49,
"height": 84
},
{
"x": 1058,
"y": 718,
"width": 1215,
"height": 84
},
{
"x": 604,
"y": 805,
"width": 399,
"height": 804
},
{
"x": 1006,
"y": 805,
"width": 49,
"height": 804
},
{
"x": 1058,
"y": 805,
"width": 1215,
"height": 804
},
{
"x": 604,
"y": 1724,
"width": 399,
"height": 84
},
{
"x": 1006,
"y": 1724,
"width": 49,
"height": 84
},
{
"x": 1058,
"y": 1724,
"width": 1215,
"height": 84
},
{
"x": 604,
"y": 1811,
"width": 399,
"height": 83
},
{
"x": 1006,
"y": 1811,
"width": 49,
"height": 83
},
{
"x": 1058,
"y": 1811,
"width": 1215,
"height": 83
},
{
"x": 604,
"y": 1897,
"width": 399,
"height": 84
},
{
"x": 1006,
"y": 1897,
"width": 49,
"height": 84
},
{
"x": 1058,
"y": 1897,
"width": 1215,
"height": 84
},
{
"x": 604,
"y": 1984,
"width": 399,
"height": 84
},
{
"x": 1006,
"y": 1984,
"width": 49,
"height": 84
},
{
"x": 1058,
"y": 1984,
"width": 1215,
"height": 84
},
{
"x": 604,
"y": 2071,
"width": 399,
"height": 813
},
{
"x": 1006,
"y": 2071,
"width": 49,
"height": 813
},
{
"x": 1058,
"y": 2071,
"width": 1215,
"height": 813
}
]
}
]
}

BIN
cv_analysis/test/test_data/test1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 241 KiB

851
cv_analysis/test/test_data/test10.json Normal file
View File

@ -0,0 +1,851 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2480,
"pageHeight": 3509,
"cells": [
{
"x": 1828,
"y": 1667,
"width": 382,
"height": 55
},
{
"x": 1477,
"y": 1670,
"width": 349,
"height": 55
},
{
"x": 1126,
"y": 1673,
"width": 349,
"height": 54
},
{
"x": 776,
"y": 1676,
"width": 348,
"height": 54
},
{
"x": 425,
"y": 1678,
"width": 348,
"height": 54
},
{
"x": 1828,
"y": 1722,
"width": 382,
"height": 55
},
{
"x": 1477,
"y": 1725,
"width": 349,
"height": 55
},
{
"x": 1126,
"y": 1728,
"width": 349,
"height": 54
},
{
"x": 776,
"y": 1730,
"width": 348,
"height": 55
},
{
"x": 425,
"y": 1733,
"width": 349,
"height": 54
},
{
"x": 1828,
"y": 1777,
"width": 382,
"height": 54
},
{
"x": 1478,
"y": 1780,
"width": 348,
"height": 54
},
{
"x": 1126,
"y": 1783,
"width": 349,
"height": 53
},
{
"x": 776,
"y": 1786,
"width": 348,
"height": 53
},
{
"x": 426,
"y": 1788,
"width": 348,
"height": 53
},
{
"x": 1828,
"y": 1832,
"width": 382,
"height": 54
},
{
"x": 1478,
"y": 1835,
"width": 348,
"height": 54
},
{
"x": 1126,
"y": 1837,
"width": 349,
"height": 55
},
{
"x": 776,
"y": 1840,
"width": 348,
"height": 54
},
{
"x": 426,
"y": 1843,
"width": 348,
"height": 53
},
{
"x": 1829,
"y": 1887,
"width": 381,
"height": 54
},
{
"x": 1478,
"y": 1890,
"width": 348,
"height": 53
},
{
"x": 1126,
"y": 1892,
"width": 349,
"height": 54
},
{
"x": 776,
"y": 1895,
"width": 348,
"height": 53
},
{
"x": 426,
"y": 1898,
"width": 348,
"height": 53
},
{
"x": 1829,
"y": 1941,
"width": 381,
"height": 54
},
{
"x": 1478,
"y": 1944,
"width": 348,
"height": 54
},
{
"x": 1126,
"y": 1947,
"width": 349,
"height": 53
},
{
"x": 776,
"y": 1949,
"width": 348,
"height": 54
},
{
"x": 426,
"y": 1952,
"width": 348,
"height": 53
},
{
"x": 1829,
"y": 1995,
"width": 381,
"height": 55
},
{
"x": 1478,
"y": 1999,
"width": 348,
"height": 53
},
{
"x": 1127,
"y": 2001,
"width": 348,
"height": 54
},
{
"x": 776,
"y": 2004,
"width": 348,
"height": 53
},
{
"x": 426,
"y": 2006,
"width": 348,
"height": 54
},
{
"x": 1829,
"y": 2050,
"width": 382,
"height": 54
},
{
"x": 1478,
"y": 2053,
"width": 348,
"height": 54
},
{
"x": 1127,
"y": 2056,
"width": 349,
"height": 54
},
{
"x": 776,
"y": 2058,
"width": 348,
"height": 54
},
{
"x": 426,
"y": 2061,
"width": 348,
"height": 54
},
{
"x": 1829,
"y": 2105,
"width": 382,
"height": 54
},
{
"x": 1478,
"y": 2108,
"width": 349,
"height": 54
},
{
"x": 1127,
"y": 2110,
"width": 349,
"height": 54
},
{
"x": 776,
"y": 2113,
"width": 348,
"height": 54
},
{
"x": 426,
"y": 2116,
"width": 348,
"height": 54
},
{
"x": 1829,
"y": 2159,
"width": 382,
"height": 55
},
{
"x": 1478,
"y": 2164,
"width": 349,
"height": 52
},
{
"x": 1127,
"y": 2166,
"width": 349,
"height": 53
},
{
"x": 776,
"y": 2169,
"width": 348,
"height": 53
},
{
"x": 426,
"y": 2172,
"width": 348,
"height": 52
},
{
"x": 1829,
"y": 2215,
"width": 382,
"height": 55
},
{
"x": 1478,
"y": 2218,
"width": 349,
"height": 54
},
{
"x": 1127,
"y": 2221,
"width": 349,
"height": 54
},
{
"x": 776,
"y": 2224,
"width": 348,
"height": 53
},
{
"x": 426,
"y": 2226,
"width": 348,
"height": 54
},
{
"x": 1830,
"y": 2271,
"width": 381,
"height": 50
},
{
"x": 1479,
"y": 2274,
"width": 348,
"height": 49
},
{
"x": 1127,
"y": 2276,
"width": 349,
"height": 50
},
{
"x": 776,
"y": 2279,
"width": 348,
"height": 49
},
{
"x": 426,
"y": 2282,
"width": 348,
"height": 49
},
{
"x": 1830,
"y": 2322,
"width": 383,
"height": 53
},
{
"x": 1480,
"y": 2325,
"width": 348,
"height": 53
},
{
"x": 1127,
"y": 2328,
"width": 350,
"height": 52
},
{
"x": 776,
"y": 2330,
"width": 348,
"height": 53
},
{
"x": 426,
"y": 2333,
"width": 348,
"height": 52
},
{
"x": 1831,
"y": 2377,
"width": 382,
"height": 52
},
{
"x": 1480,
"y": 2380,
"width": 348,
"height": 52
},
{
"x": 1127,
"y": 2382,
"width": 350,
"height": 52
},
{
"x": 776,
"y": 2385,
"width": 348,
"height": 51
},
{
"x": 426,
"y": 2388,
"width": 348,
"height": 51
},
{
"x": 1831,
"y": 2430,
"width": 382,
"height": 53
},
{
"x": 1480,
"y": 2433,
"width": 348,
"height": 53
},
{
"x": 1127,
"y": 2436,
"width": 350,
"height": 53
},
{
"x": 777,
"y": 2438,
"width": 348,
"height": 53
},
{
"x": 426,
"y": 2441,
"width": 348,
"height": 52
},
{
"x": 1831,
"y": 2485,
"width": 383,
"height": 53
},
{
"x": 1480,
"y": 2488,
"width": 348,
"height": 52
},
{
"x": 1127,
"y": 2490,
"width": 350,
"height": 53
},
{
"x": 777,
"y": 2493,
"width": 348,
"height": 52
},
{
"x": 427,
"y": 2495,
"width": 348,
"height": 53
},
{
"x": 1831,
"y": 2539,
"width": 383,
"height": 53
},
{
"x": 1480,
"y": 2542,
"width": 349,
"height": 53
},
{
"x": 1127,
"y": 2545,
"width": 350,
"height": 52
},
{
"x": 777,
"y": 2547,
"width": 348,
"height": 53
},
{
"x": 427,
"y": 2550,
"width": 348,
"height": 52
},
{
"x": 1831,
"y": 2593,
"width": 383,
"height": 54
},
{
"x": 1480,
"y": 2596,
"width": 349,
"height": 54
},
{
"x": 1127,
"y": 2599,
"width": 351,
"height": 53
},
{
"x": 777,
"y": 2601,
"width": 348,
"height": 54
},
{
"x": 427,
"y": 2604,
"width": 348,
"height": 53
},
{
"x": 1831,
"y": 2649,
"width": 383,
"height": 53
},
{
"x": 1480,
"y": 2652,
"width": 349,
"height": 52
},
{
"x": 1128,
"y": 2654,
"width": 350,
"height": 53
},
{
"x": 777,
"y": 2657,
"width": 348,
"height": 52
},
{
"x": 427,
"y": 2659,
"width": 348,
"height": 53
},
{
"x": 1832,
"y": 2703,
"width": 382,
"height": 54
},
{
"x": 1480,
"y": 2706,
"width": 349,
"height": 53
},
{
"x": 1128,
"y": 2709,
"width": 350,
"height": 53
},
{
"x": 778,
"y": 2711,
"width": 347,
"height": 53
},
{
"x": 427,
"y": 2714,
"width": 348,
"height": 53
},
{
"x": 1832,
"y": 2758,
"width": 382,
"height": 53
},
{
"x": 1481,
"y": 2761,
"width": 348,
"height": 53
},
{
"x": 1128,
"y": 2764,
"width": 350,
"height": 52
},
{
"x": 778,
"y": 2766,
"width": 348,
"height": 53
},
{
"x": 427,
"y": 2769,
"width": 348,
"height": 52
},
{
"x": 1832,
"y": 2812,
"width": 382,
"height": 55
},
{
"x": 1481,
"y": 2816,
"width": 349,
"height": 53
},
{
"x": 1128,
"y": 2818,
"width": 351,
"height": 54
},
{
"x": 778,
"y": 2820,
"width": 348,
"height": 54
},
{
"x": 428,
"y": 2823,
"width": 347,
"height": 53
},
{
"x": 1832,
"y": 2868,
"width": 382,
"height": 54
},
{
"x": 1481,
"y": 2871,
"width": 349,
"height": 53
},
{
"x": 1128,
"y": 2873,
"width": 351,
"height": 54
},
{
"x": 778,
"y": 2876,
"width": 348,
"height": 53
},
{
"x": 428,
"y": 2878,
"width": 348,
"height": 53
},
{
"x": 1832,
"y": 2923,
"width": 382,
"height": 53
},
{
"x": 1481,
"y": 2926,
"width": 349,
"height": 52
},
{
"x": 1128,
"y": 2928,
"width": 351,
"height": 53
},
{
"x": 778,
"y": 2931,
"width": 348,
"height": 52
},
{
"x": 428,
"y": 2933,
"width": 348,
"height": 52
},
{
"x": 1832,
"y": 2978,
"width": 382,
"height": 53
},
{
"x": 1481,
"y": 2980,
"width": 349,
"height": 53
},
{
"x": 1129,
"y": 2983,
"width": 350,
"height": 53
},
{
"x": 778,
"y": 2985,
"width": 348,
"height": 53
},
{
"x": 428,
"y": 2987,
"width": 348,
"height": 53
},
{
"x": 1832,
"y": 3032,
"width": 382,
"height": 53
},
{
"x": 1481,
"y": 3035,
"width": 349,
"height": 52
},
{
"x": 1129,
"y": 3038,
"width": 350,
"height": 52
},
{
"x": 779,
"y": 3040,
"width": 348,
"height": 52
},
{
"x": 428,
"y": 3042,
"width": 349,
"height": 53
},
{
"x": 1832,
"y": 3086,
"width": 382,
"height": 55
},
{
"x": 1481,
"y": 3089,
"width": 349,
"height": 54
},
{
"x": 1129,
"y": 3092,
"width": 350,
"height": 54
},
{
"x": 779,
"y": 3094,
"width": 348,
"height": 54
},
{
"x": 429,
"y": 3097,
"width": 348,
"height": 53
},
{
"x": 1832,
"y": 3141,
"width": 382,
"height": 55
},
{
"x": 1481,
"y": 3144,
"width": 349,
"height": 55
},
{
"x": 1129,
"y": 3147,
"width": 350,
"height": 54
},
{
"x": 779,
"y": 3149,
"width": 349,
"height": 55
},
{
"x": 429,
"y": 3152,
"width": 348,
"height": 54
}
]
}
]
}

BIN
cv_analysis/test/test_data/test10.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 316 KiB

839
cv_analysis/test/test_data/test2.json Normal file
View File

@ -0,0 +1,839 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2481,
"pageHeight": 3509,
"cells": [
{
"x": 327,
"y": 609,
"width": 353,
"height": 161
},
{
"x": 683,
"y": 609,
"width": 311,
"height": 161
},
{
"x": 997,
"y": 609,
"width": 531,
"height": 161
},
{
"x": 1531,
"y": 609,
"width": 247,
"height": 161
},
{
"x": 1781,
"y": 609,
"width": 246,
"height": 161
},
{
"x": 2030,
"y": 609,
"width": 246,
"height": 161
},
{
"x": 327,
"y": 773,
"width": 353,
"height": 272
},
{
"x": 683,
"y": 773,
"width": 311,
"height": 65
},
{
"x": 997,
"y": 773,
"width": 531,
"height": 65
},
{
"x": 1531,
"y": 773,
"width": 247,
"height": 65
},
{
"x": 1781,
"y": 773,
"width": 246,
"height": 65
},
{
"x": 2030,
"y": 773,
"width": 246,
"height": 65
},
{
"x": 683,
"y": 841,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 841,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 841,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 841,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 841,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 910,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 910,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 910,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 910,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 910,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 979,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 979,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 979,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 979,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 979,
"width": 246,
"height": 66
},
{
"x": 327,
"y": 1047,
"width": 353,
"height": 273
},
{
"x": 683,
"y": 1048,
"width": 311,
"height": 65
},
{
"x": 997,
"y": 1048,
"width": 531,
"height": 65
},
{
"x": 1531,
"y": 1048,
"width": 247,
"height": 65
},
{
"x": 1781,
"y": 1048,
"width": 246,
"height": 65
},
{
"x": 2030,
"y": 1048,
"width": 246,
"height": 65
},
{
"x": 683,
"y": 1116,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 1116,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 1116,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 1116,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 1116,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 1185,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 1185,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 1185,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 1185,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 1185,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 1254,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 1254,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 1254,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 1254,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 1254,
"width": 246,
"height": 66
},
{
"x": 327,
"y": 1322,
"width": 353,
"height": 273
},
{
"x": 683,
"y": 1323,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 1323,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 1323,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 1323,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 1323,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 1392,
"width": 311,
"height": 65
},
{
"x": 997,
"y": 1392,
"width": 531,
"height": 65
},
{
"x": 1531,
"y": 1392,
"width": 247,
"height": 65
},
{
"x": 1781,
"y": 1392,
"width": 246,
"height": 65
},
{
"x": 2030,
"y": 1392,
"width": 246,
"height": 65
},
{
"x": 683,
"y": 1460,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 1460,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 1460,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 1460,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 1460,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 1529,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 1529,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 1529,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 1529,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 1529,
"width": 246,
"height": 66
},
{
"x": 327,
"y": 1849,
"width": 353,
"height": 161
},
{
"x": 683,
"y": 1849,
"width": 311,
"height": 161
},
{
"x": 997,
"y": 1849,
"width": 531,
"height": 161
},
{
"x": 1531,
"y": 1849,
"width": 247,
"height": 161
},
{
"x": 1781,
"y": 1849,
"width": 246,
"height": 161
},
{
"x": 2030,
"y": 1849,
"width": 246,
"height": 161
},
{
"x": 327,
"y": 2013,
"width": 353,
"height": 272
},
{
"x": 683,
"y": 2013,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2013,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2013,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2013,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2013,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 2082,
"width": 311,
"height": 65
},
{
"x": 997,
"y": 2082,
"width": 531,
"height": 65
},
{
"x": 1531,
"y": 2082,
"width": 247,
"height": 65
},
{
"x": 1781,
"y": 2082,
"width": 246,
"height": 65
},
{
"x": 2030,
"y": 2082,
"width": 246,
"height": 65
},
{
"x": 683,
"y": 2150,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2150,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2150,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2150,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2150,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 2219,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2219,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2219,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2219,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2219,
"width": 246,
"height": 66
},
{
"x": 327,
"y": 2287,
"width": 353,
"height": 273
},
{
"x": 683,
"y": 2288,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2288,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2288,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2288,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2288,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 2357,
"width": 311,
"height": 65
},
{
"x": 997,
"y": 2357,
"width": 531,
"height": 65
},
{
"x": 1531,
"y": 2357,
"width": 247,
"height": 65
},
{
"x": 1781,
"y": 2357,
"width": 246,
"height": 65
},
{
"x": 2030,
"y": 2357,
"width": 246,
"height": 65
},
{
"x": 683,
"y": 2425,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2425,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2425,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2425,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2425,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 2494,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2494,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2494,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2494,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2494,
"width": 246,
"height": 66
},
{
"x": 327,
"y": 2562,
"width": 353,
"height": 273
},
{
"x": 683,
"y": 2563,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2563,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2563,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2563,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2563,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 2632,
"width": 311,
"height": 65
},
{
"x": 997,
"y": 2632,
"width": 531,
"height": 65
},
{
"x": 1531,
"y": 2632,
"width": 247,
"height": 65
},
{
"x": 1781,
"y": 2632,
"width": 246,
"height": 65
},
{
"x": 2030,
"y": 2632,
"width": 246,
"height": 65
},
{
"x": 683,
"y": 2700,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2700,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2700,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2700,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2700,
"width": 246,
"height": 66
},
{
"x": 683,
"y": 2769,
"width": 311,
"height": 66
},
{
"x": 997,
"y": 2769,
"width": 531,
"height": 66
},
{
"x": 1531,
"y": 2769,
"width": 247,
"height": 66
},
{
"x": 1781,
"y": 2769,
"width": 246,
"height": 66
},
{
"x": 2030,
"y": 2769,
"width": 246,
"height": 66
}
]
}
]
}

BIN
cv_analysis/test/test_data/test2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 318 KiB

233
cv_analysis/test/test_data/test3.json Normal file
View File

@ -0,0 +1,233 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2481,
"pageHeight": 3506,
"cells": [
{
"x": 195,
"y": 1519,
"width": 2091,
"height": 84
},
{
"x": 195,
"y": 1604,
"width": 583,
"height": 123
},
{
"x": 783,
"y": 1605,
"width": 1503,
"height": 124
},
{
"x": 195,
"y": 1730,
"width": 583,
"height": 65
},
{
"x": 783,
"y": 1731,
"width": 1502,
"height": 66
},
{
"x": 195,
"y": 1798,
"width": 583,
"height": 65
},
{
"x": 783,
"y": 1799,
"width": 1502,
"height": 66
},
{
"x": 195,
"y": 1866,
"width": 583,
"height": 65
},
{
"x": 782,
"y": 1867,
"width": 1503,
"height": 66
},
{
"x": 195,
"y": 1934,
"width": 583,
"height": 65
},
{
"x": 782,
"y": 1935,
"width": 1503,
"height": 66
},
{
"x": 194,
"y": 2003,
"width": 584,
"height": 64
},
{
"x": 782,
"y": 2003,
"width": 535,
"height": 66
},
{
"x": 1321,
"y": 2005,
"width": 455,
"height": 64
},
{
"x": 1780,
"y": 2005,
"width": 505,
"height": 65
},
{
"x": 193,
"y": 2071,
"width": 585,
"height": 65
},
{
"x": 782,
"y": 2071,
"width": 535,
"height": 66
},
{
"x": 1321,
"y": 2073,
"width": 455,
"height": 64
},
{
"x": 1780,
"y": 2073,
"width": 505,
"height": 65
},
{
"x": 193,
"y": 2139,
"width": 585,
"height": 65
},
{
"x": 782,
"y": 2140,
"width": 535,
"height": 65
},
{
"x": 1321,
"y": 2141,
"width": 455,
"height": 64
},
{
"x": 1780,
"y": 2141,
"width": 505,
"height": 65
},
{
"x": 193,
"y": 2207,
"width": 585,
"height": 65
},
{
"x": 782,
"y": 2208,
"width": 535,
"height": 65
},
{
"x": 1321,
"y": 2209,
"width": 455,
"height": 64
},
{
"x": 1780,
"y": 2210,
"width": 505,
"height": 64
},
{
"x": 193,
"y": 2275,
"width": 585,
"height": 66
},
{
"x": 782,
"y": 2276,
"width": 535,
"height": 65
},
{
"x": 1321,
"y": 2277,
"width": 455,
"height": 65
},
{
"x": 1780,
"y": 2278,
"width": 505,
"height": 65
},
{
"x": 193,
"y": 2343,
"width": 584,
"height": 66
},
{
"x": 782,
"y": 2344,
"width": 1503,
"height": 67
},
{
"x": 193,
"y": 2412,
"width": 584,
"height": 65
},
{
"x": 781,
"y": 2413,
"width": 1504,
"height": 66
},
{
"x": 193,
"y": 2480,
"width": 584,
"height": 65
},
{
"x": 781,
"y": 2481,
"width": 1504,
"height": 66
}
]
}
]
}

BIN
cv_analysis/test/test_data/test3.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 278 KiB

203
cv_analysis/test/test_data/test4.json Normal file
View File

@ -0,0 +1,203 @@
{
"pages": [
{
"page": 0,
"pageWidth": 3508,
"pageHeight": 2481,
"cells": [
{
"x": 299,
"y": 761,
"width": 232,
"height": 354
},
{
"x": 533,
"y": 761,
"width": 235,
"height": 354
},
{
"x": 770,
"y": 761,
"width": 205,
"height": 354
},
{
"x": 977,
"y": 761,
"width": 211,
"height": 354
},
{
"x": 1190,
"y": 761,
"width": 425,
"height": 138
},
{
"x": 1617,
"y": 761,
"width": 195,
"height": 354
},
{
"x": 1814,
"y": 761,
"width": 168,
"height": 354
},
{
"x": 1984,
"y": 761,
"width": 184,
"height": 354
},
{
"x": 2170,
"y": 761,
"width": 191,
"height": 354
},
{
"x": 2363,
"y": 761,
"width": 274,
"height": 138
},
{
"x": 2639,
"y": 761,
"width": 159,
"height": 354
},
{
"x": 2800,
"y": 761,
"width": 466,
"height": 354
},
{
"x": 1190,
"y": 901,
"width": 141,
"height": 214
},
{
"x": 1333,
"y": 901,
"width": 123,
"height": 214
},
{
"x": 1458,
"y": 901,
"width": 157,
"height": 214
},
{
"x": 2363,
"y": 901,
"width": 130,
"height": 214
},
{
"x": 2495,
"y": 901,
"width": 142,
"height": 214
},
{
"x": 299,
"y": 1121,
"width": 232,
"height": 581
},
{
"x": 533,
"y": 1121,
"width": 235,
"height": 581
},
{
"x": 770,
"y": 1121,
"width": 205,
"height": 581
},
{
"x": 977,
"y": 1121,
"width": 211,
"height": 581
},
{
"x": 1190,
"y": 1121,
"width": 141,
"height": 581
},
{
"x": 1333,
"y": 1121,
"width": 123,
"height": 581
},
{
"x": 1458,
"y": 1121,
"width": 157,
"height": 581
},
{
"x": 1617,
"y": 1121,
"width": 195,
"height": 581
},
{
"x": 1814,
"y": 1121,
"width": 168,
"height": 581
},
{
"x": 1984,
"y": 1121,
"width": 184,
"height": 581
},
{
"x": 2170,
"y": 1121,
"width": 191,
"height": 581
},
{
"x": 2363,
"y": 1121,
"width": 130,
"height": 581
},
{
"x": 2495,
"y": 1121,
"width": 142,
"height": 581
},
{
"x": 2639,
"y": 1121,
"width": 159,
"height": 581
},
{
"x": 2800,
"y": 1121,
"width": 466,
"height": 581
}
]
}
]
}

BIN
cv_analysis/test/test_data/test4.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 345 KiB

563
cv_analysis/test/test_data/test5.json Normal file
View File

@ -0,0 +1,563 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2481,
"pageHeight": 3508,
"cells": [
{
"x": 299,
"y": 706,
"width": 154,
"height": 411
},
{
"x": 455,
"y": 706,
"width": 239,
"height": 411
},
{
"x": 696,
"y": 706,
"width": 210,
"height": 411
},
{
"x": 908,
"y": 706,
"width": 818,
"height": 56
},
{
"x": 1728,
"y": 706,
"width": 376,
"height": 56
},
{
"x": 2106,
"y": 706,
"width": 138,
"height": 411
},
{
"x": 908,
"y": 764,
"width": 200,
"height": 353
},
{
"x": 1110,
"y": 764,
"width": 214,
"height": 353
},
{
"x": 1326,
"y": 764,
"width": 167,
"height": 353
},
{
"x": 1495,
"y": 764,
"width": 231,
"height": 353
},
{
"x": 1728,
"y": 764,
"width": 230,
"height": 353
},
{
"x": 1960,
"y": 764,
"width": 144,
"height": 353
},
{
"x": 299,
"y": 1123,
"width": 154,
"height": 283
},
{
"x": 455,
"y": 1123,
"width": 239,
"height": 283
},
{
"x": 696,
"y": 1123,
"width": 210,
"height": 283
},
{
"x": 908,
"y": 1123,
"width": 200,
"height": 283
},
{
"x": 1110,
"y": 1123,
"width": 214,
"height": 283
},
{
"x": 1326,
"y": 1123,
"width": 167,
"height": 283
},
{
"x": 1495,
"y": 1123,
"width": 231,
"height": 283
},
{
"x": 1728,
"y": 1123,
"width": 230,
"height": 283
},
{
"x": 1960,
"y": 1123,
"width": 144,
"height": 283
},
{
"x": 2106,
"y": 1123,
"width": 138,
"height": 283
},
{
"x": 299,
"y": 1408,
"width": 154,
"height": 284
},
{
"x": 455,
"y": 1408,
"width": 239,
"height": 284
},
{
"x": 696,
"y": 1408,
"width": 210,
"height": 284
},
{
"x": 908,
"y": 1408,
"width": 200,
"height": 284
},
{
"x": 1110,
"y": 1408,
"width": 214,
"height": 284
},
{
"x": 1326,
"y": 1408,
"width": 167,
"height": 284
},
{
"x": 1495,
"y": 1408,
"width": 231,
"height": 284
},
{
"x": 1728,
"y": 1408,
"width": 230,
"height": 284
},
{
"x": 1960,
"y": 1408,
"width": 144,
"height": 284
},
{
"x": 2106,
"y": 1408,
"width": 138,
"height": 284
},
{
"x": 299,
"y": 2090,
"width": 169,
"height": 211
},
{
"x": 470,
"y": 2090,
"width": 253,
"height": 211
},
{
"x": 725,
"y": 2090,
"width": 229,
"height": 211
},
{
"x": 956,
"y": 2090,
"width": 516,
"height": 57
},
{
"x": 1474,
"y": 2090,
"width": 496,
"height": 57
},
{
"x": 1972,
"y": 2090,
"width": 272,
"height": 211
},
{
"x": 956,
"y": 2149,
"width": 184,
"height": 152
},
{
"x": 1142,
"y": 2149,
"width": 158,
"height": 152
},
{
"x": 1302,
"y": 2149,
"width": 170,
"height": 152
},
{
"x": 1474,
"y": 2149,
"width": 262,
"height": 152
},
{
"x": 1738,
"y": 2149,
"width": 232,
"height": 152
},
{
"x": 299,
"y": 2303,
"width": 169,
"height": 56
},
{
"x": 470,
"y": 2303,
"width": 253,
"height": 56
},
{
"x": 725,
"y": 2303,
"width": 229,
"height": 56
},
{
"x": 956,
"y": 2303,
"width": 184,
"height": 56
},
{
"x": 1142,
"y": 2303,
"width": 158,
"height": 56
},
{
"x": 1302,
"y": 2303,
"width": 170,
"height": 56
},
{
"x": 1474,
"y": 2303,
"width": 262,
"height": 56
},
{
"x": 1738,
"y": 2303,
"width": 232,
"height": 56
},
{
"x": 1972,
"y": 2303,
"width": 272,
"height": 56
},
{
"x": 299,
"y": 2361,
"width": 169,
"height": 204
},
{
"x": 470,
"y": 2361,
"width": 253,
"height": 204
},
{
"x": 725,
"y": 2361,
"width": 229,
"height": 97
},
{
"x": 956,
"y": 2361,
"width": 184,
"height": 97
},
{
"x": 1142,
"y": 2361,
"width": 158,
"height": 204
},
{
"x": 1302,
"y": 2361,
"width": 170,
"height": 204
},
{
"x": 1474,
"y": 2361,
"width": 262,
"height": 97
},
{
"x": 1738,
"y": 2361,
"width": 232,
"height": 97
},
{
"x": 1972,
"y": 2361,
"width": 272,
"height": 204
},
{
"x": 725,
"y": 2460,
"width": 229,
"height": 105
},
{
"x": 956,
"y": 2460,
"width": 184,
"height": 105
},
{
"x": 1474,
"y": 2460,
"width": 262,
"height": 105
},
{
"x": 1738,
"y": 2460,
"width": 232,
"height": 105
},
{
"x": 299,
"y": 2567,
"width": 169,
"height": 205
},
{
"x": 470,
"y": 2567,
"width": 253,
"height": 205
},
{
"x": 725,
"y": 2567,
"width": 229,
"height": 205
},
{
"x": 956,
"y": 2567,
"width": 184,
"height": 205
},
{
"x": 1142,
"y": 2567,
"width": 158,
"height": 205
},
{
"x": 1302,
"y": 2567,
"width": 170,
"height": 205
},
{
"x": 1474,
"y": 2567,
"width": 262,
"height": 205
},
{
"x": 1738,
"y": 2567,
"width": 232,
"height": 205
},
{
"x": 1972,
"y": 2567,
"width": 272,
"height": 205
},
{
"x": 299,
"y": 2774,
"width": 169,
"height": 56
},
{
"x": 470,
"y": 2774,
"width": 253,
"height": 56
},
{
"x": 725,
"y": 2774,
"width": 229,
"height": 56
},
{
"x": 956,
"y": 2774,
"width": 184,
"height": 56
},
{
"x": 1142,
"y": 2774,
"width": 158,
"height": 56
},
{
"x": 1302,
"y": 2774,
"width": 170,
"height": 56
},
{
"x": 1474,
"y": 2774,
"width": 262,
"height": 56
},
{
"x": 1738,
"y": 2774,
"width": 232,
"height": 56
},
{
"x": 1972,
"y": 2774,
"width": 272,
"height": 56
},
{
"x": 299,
"y": 2832,
"width": 169,
"height": 205
},
{
"x": 470,
"y": 2832,
"width": 253,
"height": 205
},
{
"x": 725,
"y": 2832,
"width": 229,
"height": 205
},
{
"x": 956,
"y": 2832,
"width": 184,
"height": 205
},
{
"x": 1142,
"y": 2832,
"width": 158,
"height": 205
},
{
"x": 1302,
"y": 2832,
"width": 170,
"height": 205
},
{
"x": 1474,
"y": 2832,
"width": 262,
"height": 205
},
{
"x": 1738,
"y": 2832,
"width": 232,
"height": 205
},
{
"x": 1972,
"y": 2832,
"width": 272,
"height": 205
}
]
}
]
}

BIN
cv_analysis/test/test_data/test5.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 557 KiB

53
cv_analysis/test/test_data/test6.json Normal file
View File

@ -0,0 +1,53 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2480,
"pageHeight": 3509,
"cells": [
{
"x": 494,
"y": 960,
"width": 562,
"height": 453
},
{
"x": 1059,
"y": 962,
"width": 1065,
"height": 224
},
{
"x": 1060,
"y": 1190,
"width": 1066,
"height": 225
},
{
"x": 500,
"y": 1419,
"width": 559,
"height": 521
},
{
"x": 1060,
"y": 1420,
"width": 1070,
"height": 521
},
{
"x": 506,
"y": 1945,
"width": 553,
"height": 222
},
{
"x": 1063,
"y": 1946,
"width": 1070,
"height": 223
}
]
}
]
}

BIN
cv_analysis/test/test_data/test6.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 95 KiB

29
cv_analysis/test/test_data/test7.json Normal file
View File

@ -0,0 +1,29 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2471,
"pageHeight": 3505,
"cells": [
{
"x": 572,
"y": 725,
"width": 451,
"height": 1785
},
{
"x": 1025,
"y": 725,
"width": 505,
"height": 1785
},
{
"x": 1533,
"y": 724,
"width": 454,
"height": 1786
}
]
}
]
}

BIN
cv_analysis/test/test_data/test7.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 142 KiB

179
cv_analysis/test/test_data/test8.json Normal file
View File

@ -0,0 +1,179 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2471,
"pageHeight": 3505,
"cells": [
{
"x": 1292,
"y": 1009,
"width": 264,
"height": 132
},
{
"x": 538,
"y": 1015,
"width": 254,
"height": 124
},
{
"x": 797,
"y": 1016,
"width": 253,
"height": 124
},
{
"x": 1056,
"y": 1017,
"width": 230,
"height": 123
},
{
"x": 538,
"y": 1144,
"width": 253,
"height": 81
},
{
"x": 797,
"y": 1145,
"width": 253,
"height": 82
},
{
"x": 1056,
"y": 1146,
"width": 230,
"height": 81
},
{
"x": 1292,
"y": 1146,
"width": 257,
"height": 81
},
{
"x": 538,
"y": 1231,
"width": 253,
"height": 39
},
{
"x": 797,
"y": 1232,
"width": 253,
"height": 39
},
{
"x": 1055,
"y": 1233,
"width": 230,
"height": 38
},
{
"x": 1291,
"y": 1233,
"width": 258,
"height": 38
},
{
"x": 538,
"y": 1277,
"width": 253,
"height": 80
},
{
"x": 797,
"y": 1277,
"width": 252,
"height": 80
},
{
"x": 1055,
"y": 1278,
"width": 230,
"height": 80
},
{
"x": 1291,
"y": 1278,
"width": 258,
"height": 80
},
{
"x": 538,
"y": 1362,
"width": 253,
"height": 40
},
{
"x": 797,
"y": 1363,
"width": 253,
"height": 40
},
{
"x": 1055,
"y": 1363,
"width": 230,
"height": 40
},
{
"x": 1291,
"y": 1363,
"width": 258,
"height": 40
},
{
"x": 538,
"y": 1407,
"width": 253,
"height": 82
},
{
"x": 797,
"y": 1408,
"width": 253,
"height": 81
},
{
"x": 1055,
"y": 1408,
"width": 231,
"height": 82
},
{
"x": 1291,
"y": 1409,
"width": 258,
"height": 81
},
{
"x": 538,
"y": 1494,
"width": 254,
"height": 209
},
{
"x": 797,
"y": 1494,
"width": 253,
"height": 209
},
{
"x": 1055,
"y": 1495,
"width": 231,
"height": 209
},
{
"x": 1291,
"y": 1495,
"width": 265,
"height": 214
}
]
}
]
}

BIN
cv_analysis/test/test_data/test8.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 76 KiB

335
cv_analysis/test/test_data/test9.json Normal file
View File

@ -0,0 +1,335 @@
{
"pages": [
{
"page": 0,
"pageWidth": 2481,
"pageHeight": 3506,
"cells": [
{
"x": 1913,
"y": 602,
"width": 56,
"height": 670
},
{
"x": 1973,
"y": 602,
"width": 56,
"height": 670
},
{
"x": 2032,
"y": 603,
"width": 57,
"height": 669
},
{
"x": 605,
"y": 647,
"width": 389,
"height": 430
},
{
"x": 997,
"y": 646,
"width": 210,
"height": 432
},
{
"x": 1211,
"y": 647,
"width": 210,
"height": 432
},
{
"x": 1425,
"y": 647,
"width": 66,
"height": 432
},
{
"x": 1496,
"y": 647,
"width": 65,
"height": 432
},
{
"x": 605,
"y": 1083,
"width": 388,
"height": 350
},
{
"x": 997,
"y": 1083,
"width": 209,
"height": 352
},
{
"x": 1210,
"y": 1083,
"width": 211,
"height": 352
},
{
"x": 1424,
"y": 1083,
"width": 67,
"height": 352
},
{
"x": 1495,
"y": 1083,
"width": 66,
"height": 352
},
{
"x": 1912,
"y": 1275,
"width": 57,
"height": 669
},
{
"x": 1972,
"y": 1275,
"width": 57,
"height": 670
},
{
"x": 2032,
"y": 1276,
"width": 57,
"height": 669
},
{
"x": 604,
"y": 1439,
"width": 389,
"height": 498
},
{
"x": 996,
"y": 1439,
"width": 67,
"height": 499
},
{
"x": 1067,
"y": 1439,
"width": 67,
"height": 499
},
{
"x": 1138,
"y": 1439,
"width": 68,
"height": 500
},
{
"x": 1210,
"y": 1439,
"width": 67,
"height": 500
},
{
"x": 1280,
"y": 1439,
"width": 67,
"height": 500
},
{
"x": 1351,
"y": 1439,
"width": 69,
"height": 500
},
{
"x": 1424,
"y": 1439,
"width": 67,
"height": 500
},
{
"x": 1495,
"y": 1439,
"width": 65,
"height": 500
},
{
"x": 603,
"y": 1943,
"width": 389,
"height": 291
},
{
"x": 996,
"y": 1943,
"width": 209,
"height": 292
},
{
"x": 1209,
"y": 1943,
"width": 210,
"height": 292
},
{
"x": 1424,
"y": 1943,
"width": 67,
"height": 292
},
{
"x": 1494,
"y": 1943,
"width": 66,
"height": 293
},
{
"x": 1911,
"y": 1948,
"width": 56,
"height": 669
},
{
"x": 1971,
"y": 1948,
"width": 56,
"height": 669
},
{
"x": 2030,
"y": 1949,
"width": 57,
"height": 669
},
{
"x": 603,
"y": 2239,
"width": 388,
"height": 304
},
{
"x": 995,
"y": 2239,
"width": 67,
"height": 305
},
{
"x": 1066,
"y": 2239,
"width": 67,
"height": 306
},
{
"x": 1137,
"y": 2239,
"width": 68,
"height": 306
},
{
"x": 1209,
"y": 2239,
"width": 66,
"height": 306
},
{
"x": 1280,
"y": 2240,
"width": 67,
"height": 305
},
{
"x": 1351,
"y": 2240,
"width": 68,
"height": 305
},
{
"x": 1423,
"y": 2240,
"width": 67,
"height": 305
},
{
"x": 1494,
"y": 2240,
"width": 65,
"height": 305
},
{
"x": 601,
"y": 2548,
"width": 390,
"height": 783
},
{
"x": 995,
"y": 2548,
"width": 66,
"height": 783
},
{
"x": 1065,
"y": 2548,
"width": 68,
"height": 783
},
{
"x": 1137,
"y": 2549,
"width": 68,
"height": 782
},
{
"x": 1209,
"y": 2549,
"width": 66,
"height": 782
},
{
"x": 1279,
"y": 2549,
"width": 67,
"height": 782
},
{
"x": 1350,
"y": 2549,
"width": 69,
"height": 782
},
{
"x": 1423,
"y": 2549,
"width": 67,
"height": 782
},
{
"x": 1493,
"y": 2549,
"width": 66,
"height": 782
},
{
"x": 1910,
"y": 2622,
"width": 57,
"height": 666
},
{
"x": 1970,
"y": 2622,
"width": 57,
"height": 667
},
{
"x": 2030,
"y": 2622,
"width": 56,
"height": 667
}
]
}
]
}

BIN
cv_analysis/test/test_data/test9.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 334 KiB

32
cv_analysis/test/unit_tests/table_test.py
View File

@ -1,7 +1,7 @@
from os.path import join
import json
from cv_analysis.table_parsing import parse_table
from cv_analysis.table_parsing import parse_tables
from cv_analysis.locations import TEST_DATA_DIR
from cv_analysis.test.config import TEST_CONFIG
from cv_analysis.utils.test_metrics import compute_document_score
@ -9,13 +9,37 @@ from cv_analysis.utils.preprocessing import open_pdf
def test_table_parsing():
for i in range(1, 11):
img_path = join(TEST_DATA_DIR, f"test{i}.png")
json_path = join(TEST_DATA_DIR, f"test{i}.json")
pages = open_pdf(img_path)
result = {"pages": []}
for i, page in enumerate(pages):
result["pages"].append({"page": str(i), "cells": [x.json_xywh() for x in parse_tables(page)]})
with open(json_path) as f:
annotation = json.load(f)
score = compute_document_score(result, annotation)
assert round(score, 3) >= TEST_CONFIG.table_score_threshold
"""
def test_table_parsing():
img_path = join(TEST_DATA_DIR, "table.jpg")
json_path = join(TEST_DATA_DIR, "table.json")
pages = open_pdf(img_path)[0]
result = {}
pages = open_pdf(img_path)
result = {"pages": []}
for i, page in enumerate(pages):
result.update({str(i): parse_table(page)})
result["pages"].append({"page": str(i), "cells": [x.xywh() for x in parse_tables(page)]})
with open(json_path) as f:
annotation = json.load(f)
score = compute_document_score(result, annotation)
assert score >= TEST_CONFIG.table_score_threshold
"""

3
cv_analysis/utils/draw.py
View File

@ -3,7 +3,7 @@ import cv2
from cv_analysis.utils import copy_and_normalize_channels
def draw_contours(image, contours):
def draw_contours(image, contours, color=None, annotate=False):
image = copy_and_normalize_channels(image)
@ -25,7 +25,6 @@ def draw_rectangles(image, rectangles, color=None, annotate=False):
for rect in rectangles:
x, y, w, h = rect
cv2.rectangle(image, (x, y), (x + w, y + h), color, 2)
if annotate:
annotate_rect(x, y, w, h)

16
cv_analysis/utils/preprocessing.py
View File

@ -3,29 +3,29 @@ import pdf2image
from PIL import Image
import cv2
from cv_analysis.utils.deskew import deskew
def preprocess_pdf_image(page):
if len(page.shape) > 2:
page = cv2.cvtColor(page, cv2.COLOR_BGR2GRAY)
page = cv2.fastNlMeansDenoising(page, h=3)
return deskew(page)
return page
def open_pdf(pdf, first_page=0, last_page=None):
first_page += 1
last_page = None if last_page is None else last_page + 1
if type(pdf) == str:
if pdf.endswith(".jpg") or pdf.endswith(".png"):
if pdf.lower().endswith((".png", ".jpg", ".jpeg")):
pages = [Image.open(pdf)]
# assume pdf as default file type for a path argument
else:
else: # assume pdf as default file type for a path argument
pages = pdf2image.convert_from_path(pdf, first_page=first_page, last_page=last_page)
elif type(pdf) == bytes:
pages = pdf2image.convert_from_bytes(pdf, first_page=first_page, last_page=last_page)
elif type(pdf) in {list, ndarray}:
return pdf
pages = [preprocess_pdf_image(array(p)) for p in pages]
pages, angles = list(zip(*pages))
return pages, angles
return pages

70
cv_analysis/utils/structures.py Normal file
View File

@ -0,0 +1,70 @@
from json import dumps
class Rectangle:
def __init__(self, x1=None, y1=None, w=None, h=None, x2=None, y2=None, indent=4, format="xywh"):
try:
self.x1 = x1
self.y1 = y1
self.w = w if w else x2 - x1
self.h = h if h else y2 - y1
self.x2 = x2 if x2 else x1 + w
self.y2 = y2 if y2 else y1 + h
assert (self.x1 + self.w) == self.x2
assert (self.y1 + self.h) == self.y2
self.indent = indent
self.format = format
except:
raise Exception("x1, y1, (w|x2), and (h|y2) must be defined.")
def json_xywh(self):
return {"x": self.x1, "y": self.y1, "width": self.w, "height": self.h}
def json_xyxy(self):
return {"x1": self.x1, "y1": self.y1, "x2": self.x2, "y2": self.y2}
def json_full(self):
return {"x1": self.x1, "y1": self.y1, "x2": self.x2, "y2": self.y2, "width": self.w, "height": self.h}
def json(self):
json_func = {"xywh": self.json_xywh, "xyxy": self.json_xyxy}.get(self.format, self.json_full)
return json_func()
def xyxy(self):
return self.x1, self.y1, self.x2, self.y2
def xywh(self):
return self.x1, self.y1, self.w, self.h
@classmethod
def from_xyxy(cls, xyxy_tuple):
x1, y1, x2, y2 = xyxy_tuple
return cls(x1=x1, y1=y1, x2=x2, y2=y2)
@classmethod
def from_xywh(cls, xywh_tuple):
x, y, w, h = xywh_tuple
return cls(x1=x, y1=y, w=w, h=h)
def __str__(self):
return dumps(self.json(), indent=self.indent)
def __repr__(self):
return str(self.json())
def __iter__(self):
return list(self.json().values()).__iter__()
"""
boxes = [[30,40,5,6],[56,78,23,19],[5,100,45,35],[34,34,67,67]]
rectangles = list(map(Rectangle.from_xywh, boxes))
rectangles
r = rectangles[1]
"""
class Contour:
def __init__(self):
pass

49
cv_analysis/utils/test_metrics.py
View File

@ -1,12 +1,41 @@
import numpy as np
from cv_analysis.utils.structures import Rectangle
def compute_iou_from_boxes(box1, box2):
def xyxy_from_object(box_object):
try:
x1, y1, x2, y2 = box_object.xyxy()
except:
try:
x1 = box_object["x"]
y1 = box_object["y"]
x2 = x1 + box_object["width"]
y2 = y1 + box_object["height"]
except:
x1, y1, x2, y2 = box_object
return x1, y1, x2, y2
def xywh_from_object(box_object):
try:
x, y, w, h = box_object.xywh()
except:
try:
x = box_object["x"]
y = box_object["y"]
w = box_object["width"]
h = box_object["height"]
except:
x, y, w, h = box_object
return x, y, w, h
def compute_iou_from_boxes(box1: Rectangle, box2: list):
"""
Each box of the form (x1, y1, delx, dely)
"""
ax1, ay1, aw, ah = box1
bx1, by1, bw, bh = box2
ax1, ay1, aw, ah = xywh_from_object(box1)
bx1, by1, bw, bh = xywh_from_object(box2)
ax2, ay2, bx2, by2 = ax1 + aw, ay1 + ah, bx1 + bw, by1 + bh
if (ax1 > bx2) or (bx1 > ax2) or (ay1 > by2) or (by1 > ay2):
return 0
@ -40,15 +69,15 @@ def compute_page_iou(results_box_list, gt_box_list):
def compute_document_score(results_dict, annotation_dict):
page_weights = np.array([len(v) for v in annotation_dict.values()])
page_weights = np.array([len(page["cells"]) for page in annotation_dict["pages"]])
page_weights = page_weights / sum(page_weights)
scores = []
for key in annotation_dict:
scores.append(compute_page_iou(results_dict[key], annotation_dict[key]))
for i in range(len(annotation_dict["pages"])):
scores.append(compute_page_iou(results_dict["pages"][i]["cells"], annotation_dict["pages"][i]["cells"]))
scores = np.array(scores)
doc_score = np.average(scores, weights=page_weights)
return doc_score
def compute_document_score_tables(results_dict, annotation_dict):
pass

29
cv_analysis/utils/visual_logging.py
View File

@ -10,13 +10,30 @@ class VisualLogger:
if not os.path.exists(self.output_folder):
os.mkdir(self.output_folder)
def debug(self, img, name):
if self.level_is_debug():
output_path = os.path.join(self.output_folder, name)
save_mpl(img, output_path)
def _save(self, img, name):
output_path = os.path.join(self.output_folder, name)
save_mpl(img, output_path)
def level_is_debug(self):
return self.level == "DEBUG"
def info(self, img, name):
if self._level_is_info():
self._save(img, name)
def debug(self, img, name):
if self._level_is_debug():
self._save(img, name)
def all(self, img, name):
if self._level_is_debug():
self._save(img, name)
def _level_is_info(self):
return self.level in {"INFO", "DEBUG", "ALL"}
def _level_is_debug(self):
return self.level in {"DEBUG", "ALL"}
def _level_is_all(self):
return self.level == "ALL"
vizlogger = VisualLogger(CONFIG.visual_logging.level, CONFIG.visual_logging.output_folder)

42
scripts/annotate.py
View File

@ -1,30 +1,48 @@
"""
Usage:
python scripts/annotate.py /home/iriley/Documents/pdf/scanned/10.pdf 5 --type table --show
python scripts/annotate.py /home/iriley/Documents/pdf/scanned/10.pdf 5 --type redaction --show
python scripts/annotate.py /home/iriley/Documents/pdf/scanned/10.pdf 5 --type layout --show
python scripts/annotate.py /home/iriley/Documents/pdf/scanned/10.pdf 5 --type figure --show
"""
import argparse
from cv_analysis.table_parsing import annotate_tables_in_pdf
from cv_analysis.redaction_detection import annotate_redactions_in_pdf
from cv_analysis.layout_parsing import annotate_layout_in_pdf
from cv_analysis.figure_detection import detect_figures_in_pdf
from cv_analysis.utils.display import show_mpl
from cv_analysis.utils.draw import draw_contours, draw_rectangles
from cv_analysis.utils.preprocessing import open_pdf
from cv_analysis.utils.visual_logging import vizlogger
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument("pdf_path")
parser.add_argument("page_index", type=int)
parser.add_argument("--type", choices=["table", "redaction", "layout", "figure", "figures"])
parser.add_argument("--page_index", type=int, default=0)
parser.add_argument("--type", choices=["table", "redaction", "layout", "figure"], default="table")
parser.add_argument("--show", action="store_true", default=False)
args = parser.parse_args()
return args
def annotate_page(page_image, analysis_function, drawing_function, name="tmp.png", show=True):
result = analysis_function(page_image)
page_image = drawing_function(page_image, result)
vizlogger.debug(page_image, "redactions05_output.png")
show_mpl(page_image)
if __name__ == "__main__":
args = parse_args()
page = open_pdf(args.pdf_path, first_page=args.page_index, last_page=args.page_index)[0]
name = f"{args.type}_final_result.png"
draw = draw_rectangles
if args.type == "table":
annotate_tables_in_pdf(args.pdf_path, page_index=args.page_index, show=args.show)
from cv_analysis.table_parsing import parse_tables as analyze
elif args.type == "redaction":
annotate_redactions_in_pdf(args.pdf_path, page_index=args.page_index, show=args.show)
from cv_analysis.redaction_detection import find_redactions as analyze
draw = draw_contours
elif args.type == "layout":
annotate_layout_in_pdf(args.pdf_path, page_index=args.page_index, show=args.show)
from cv_analysis.layout_parsing import parse_layout as analyze
elif args.type == "figure":
detect_figures_in_pdf(args.pdf_path, page_index=args.page_index, show=True)
from cv_analysis.figure_detection import detect_figures as analyze
annotate_page(page, analyze, draw, name=name, show=args.show)

6
scripts/deskew_demo.py
View File

@ -7,7 +7,7 @@ from PIL import Image
from cv_analysis.utils.deskew import deskew_histbased # , deskew_linebased
from cv_analysis.utils.display import show_mpl
from cv_analysis.utils.draw import draw_stats
from cv_analysis.table_parsing import parse_table
from cv_analysis.table_parsing import parse_tables
def parse_args():
@ -37,10 +37,10 @@ if __name__ == "__main__":
page_corr_ = Image.fromarray(page_corr).convert("RGB")
page_corr_.save(args.save_path.replace(".pdf", "_corrected.pdf"))
# annotate_tables_in_pdf(args.pdf_path, page_index=args.page_index)
stats = parse_table(page)
stats = parse_tables(page)
page = draw_stats(page, stats)
show_mpl(page)
stats_corr = parse_table(page_corr)
stats_corr = parse_tables(page_corr)
page_corr = draw_stats(page_corr, stats_corr)
show_mpl(page_corr)
if args.save_path:

0
cv_analysis/test/scripts/export_example_pages.py → scripts/export_example_pages.py
View File

12
src/run_service.py
View File

@ -8,12 +8,14 @@ from prometheus_flask_exporter import PrometheusMetrics
from waitress import serve
from cv_analysis.utils import npconvert
from cv_analysis.table_parsing import parse_table
from cv_analysis.table_parsing import parse_tables
from cv_analysis.redaction_detection import find_redactions
from cv_analysis.layout_parsing import parse_layout
from cv_analysis.figure_detection import detect_figures
from cv_analysis.utils.logging import logger
from cv_analysis.utils.post_processing import Rectangle
from cv_analysis.utils.preprocessing import open_pdf
from cv_analysis.utils.structures import Rectangle
from cv_analysis.config import CONFIG
@ -42,7 +44,7 @@ def main():
@metrics.summary("tables_request_time_seconds", "Time spent processing tables request")
def get_tables():
start_monitoring()
tables = annotate(parse_table)
tables = annotate(parse_tables)
return tables
@app.route("/redactions", methods=["POST"])
@ -86,7 +88,7 @@ def make_annotations(pdf, annotation_function):
results = []
for i, page in enumerate(pdf):
boxes = annotation_function(page)
cells = [{"x": x, "y": y, "width": w, "height": h} for x, y, w, h in boxes]
cells = list(map(lambda x: x.json_xywh(), boxes))
results.append({"page": i, "pageWidth": page.shape[1], "pageHeight": page.shape[0], "cells": cells})
output_dict = {"pages": results}
return jsonify(json.dumps(output_dict, default=npconvert))
@ -101,10 +103,8 @@ def annotate(annotation_function):
data = request.data
logger.info(f"Received data.")
logger.info(f"Processing data.")
pdf, angles = open_pdf(data)
pdf = open_pdf(data)
annotations = make_annotations(pdf, annotation_function)
# if CONFIG.deskew.function != "identity":
# annotations.update({"deskew_angles": angles})
return annotations
try: