This is a fork of the Pi-Tag fiducial marker detector in the
cob_fiducials ROS
module, modified to not require ROS. Other modifications include a separate
function to just get the marker pixel coordinates instead of the pose and the
addition of the ellipse refinement step
mentioned in the Pi-Tag paper.
One example marker is included in the markers directory, copied from
cob_fiducials. Others can be generated with a
Pi-Tag generator.
The included test program can be invoked with ./pitag-test modelfile imagefile,
which will print a list of detected tags and the image pixel locations of their
dots.
- LGPL v3+ (per the
cob_fiducialslicense) - The ellipse refinement step, copied from the RUNE-Tag reference implementation, and the test program are licensed under the MIT license
- Original detector implementation:
cobs_fiducials - Ellipse refinement: RUNE-Tag reference implementation
- Pi-Tag paper: Bergamasco, Filippo, Andrea Albarelli, and Andrea Torsello. "Pi-tag: a fast image-space marker design based on projective invariants." Machine vision and applications 24, no. 6 (2013): 1295-1310.
- This standalone detector: Matthew Petroff