Runtime dependencies

Slim:

• python3

Normal (in addition to the previous):

• libopencv-imgcodecs (Ubuntu 18.04/20.04) | libopencv-highgui (Ubuntu 16.04)

• libopencv-imgproc

• libopencv-core

• libgdal

Building dependencies

Slim:

• python3-setuptools

• python3-pkgconfig

• python3-wheel

• python3-dev

• swig

Normal (in addition to the previous):

• libopencv-dev

• libgdal-dev

• libboost-all-dev

Create a new Page XML adding regions, text and properties

import pagexml
pxml = pagexml.PageXML()

# Create a new page xml
file = 'example_image.jpg'
width = 400
height = 200
pxml.newXml('name-and-version-of-tool', file, width, height)

# Add a text region to the Page
page = pxml.selectNth('//_:Page', 0)
reg = pxml.addTextRegion(page)

# Set text region bounding box with a confidence
pxml.setCoordsBBox(reg, 10, 20, 80, 60, 0.8)

# Set the text for the text region with a confidence
pxml.setTextEquiv(reg, 'lorem ipsum', 0.9)

# Add property to text region
pxml.setProperty(reg, 'key', 'value')

# Add a second page with a text region and specific id
page = pxml.addPage('example_image_2.jpg', 300, 300)
reg = pxml.addTextRegion(page, 'regA')
pxml.setCoordsBBox(reg, 15, 12, 76, 128)

# Write XML to file
pxml.write('example_image.xml')

Modify an existing Page XML

# Load an existing XML
import pagexml
pxml = pagexml.PageXML('example_image.xml')

# Add content to loaded XML
pxml.setProperty(pxml.selectNth('//_:Page', 0), 'key', 'value')

# Write XML to file
pxml.write('example_image_2.xml')

Crop an element and save image to disk

# Load an existing XML
import pagexml
pxml = pagexml.PageXML('examples/lorem.xml')

# Crop element with specific ID
cropped = pxml.crop('//*[@id="r1_l1"]/_:Coords')[0]

# Save image to disk
pagexml.imwrite(cropped.name+'.png', cropped.image)