author	Daniel Mohr
license	LGPL-3.0-or-later
home	https://gitlab.com/ug-cp/detloclcheck
mirror	https://github.com/ug-cp/detloclcheck
documentation	https://ug-cp.gitlab.io/detloclcheck
latest_release	https://gitlab.com/ug-cp/detloclcheck/-/releases/permalink/latest
doi	https://doi.org/10.5281/zenodo.15002509

DetLocLCheck -- Detection, Localization, Checkerboard, L marker

intro

DetLocLCheck is a software tool designed for the Detection and Localization of Checkerboard calibration targets containing L-shape markers. This tool utilizes template matching for initial detection, followed by refinement using OpenCV's cornerSubPix function to achieve subpixel accuracy. Finally, world coordinates are assigned to the detected markers.

install

Before installing DetLocLCheck, ensure that the required dependencies are met:

We need at least the following Debian packages:

• python3-numpy (numpy.org)
• python3-opencv (opencv.org)
• python3-pip (pip.pypa.io)

Important: On Ubuntu 22.04, do not install python3-hatchling and python3-pathspec as they are too old.

Note that numpy is defined as a dependency in pyproject.toml, but we strongly recommend using the package from your operating system's package management system.

Similarly, opencv-python is not defined as a dependency in pyproject.toml. The reason is that pip ignores the package from the package management system from the operating system. As before we strongly recommend to use the package from the package management system of your operating system.

If you still want to install opencv-python using pip, use the following command (not recommended!):

pip3 install opencv-python

The recommended way to install DetLocLCheck is:

pip3 install --break-system-packages --user https://gitlab.com/ug-cp/detloclcheck/-/archive/0.1.2/detloclcheck-0.1.2.zip

On Ubuntu 22.04 the new flag --break-system-packages is not available and you should do:

pip3 install --user https://gitlab.com/ug-cp/detloclcheck/-/archive/0.1.2/detloclcheck-0.1.2.zip

For development you could install an editable version, e. g.:

pip3 install --break-system-packages -e https://gitlab.com/ug-cp/detloclcheck/-/archive/0.1.2/detloclcheck-0.1.2.zip

This method only works with Python 3.10 and later.

If you really want to use the latest packages via pip you can do:

python3 -m venv --upgrade-deps "$HOME"/.venv_DetLocLCheck
. "$HOME"/.venv_DetLocLCheck/bin/activate
pip3 install --user "DetLocLCheck[opencv-python] @ git+https://gitlab.com/ug-cp/detloclcheck.git"

Or for the headless version of opencv:

pip3 install --user "DetLocLCheck[opencv-python-headless] @ git+https://gitlab.com/ug-cp/detloclcheck.git"

Example

Here are some examples of using DetLocLCheck:

Find and detect checkerboards in multiple images and handle the images in parallel:

detloclcheck find_checkerboard -log_file cam.log -run_parallel \
    -crosssizes 35 55 -min_sharpness 25 50 100 -f *.png

Create example data, do the detection, and visualize the result:

detloclcheck create_checkerboard_image -outfile foo.png -size 23
detloclcheck find_checkerboard -f foo.png
detloclcheck visualize foo.json -i foo.png

You can also use the Python module detloclcheck instead of the command-line interface:

import cv2

import matplotlib.pyplot

from detloclcheck.detect_localize_checkerboard import \
    detect_localize_checkerboard

image_name = 'foo.png'
gray_image = cv2.imread(image_name, cv2.COLOR_BGR2GRAY)
coordinate_system, zeropoint, axis1, axis2 = \
    detect_localize_checkerboard(
        gray_image, (11, 23),
        (0.0,  22.5,  45.0,  67.5,  90.0, 112.5, 135.0, 157.5))
matplotlib.pyplot.imshow(gray_image, cmap="Greys")
matplotlib.pyplot.plot(
    coordinate_system[:, 0, 0], coordinate_system[:, 0, 1],
    'r2', markersize=20)
matplotlib.pyplot.plot(zeropoint[0], zeropoint[1], 'b1', markersize=20)
for i in range(coordinate_system.shape[0]):
    matplotlib.pyplot.text(
        coordinate_system[i, 0, 0], coordinate_system[i, 0, 1],
        f'({int(coordinate_system[i, 1, 0])},'
        f'{int(coordinate_system[i, 1, 1])})',
        color='g', fontsize='small', rotation=45)
matplotlib.pyplot.show()

Citation

If you are using detloclcheck, please make it clear by citing:

Mohr, D. (2025). DetLocLCheck. Zenodo. https://doi.org/10.5281/zenodo.15002509

If you are using a specific version, please use the corresponding DOI.

copyright + license

Author: Daniel Mohr.

Date: 2025-07-16 (last change).

License: LGPL-3.0-or-later

Copyright (C) 2024-2025 Daniel Mohr