Package Summary

Tags No category tags.
Version 0.10.0
License BSD
Build type CMAKE

Repository Summary

Checkout URI
VCS Type git
VCS Version master
Last Updated 2019-02-24
CI status No Continuous Integration
Tags No category tags.
Contributing Help Wanted (0)
Good First Issues (0)
Pull Requests to Review (0)

Package Description

AprilTag detector library

Additional Links


  • Christian Rauch


  • Edwin Olson

AprilTag 3

AprilTags are a visual fiducial system popular in robotics research. This repository contains the most recent version of AprilTags, AprilTag 3, which includes a faster (>2x) detector, improved detection rate on small tags, flexible tag layouts, and pose estimation.

You can find tag images for the pre-generated layouts here.


Please create an issue on this github for any questions instead of sending a private message. This allows other people with the same question to find your answer.

Flexible Layouts

AprilTag 3 supports a wide variety of possible tag layouts in addition to the classic layout supported in AprilTag 2. The tag's data bits can now go outside of the tag border, and it is also possible to define layouts with "holes" inside of the tag border where there are no data bits. In this repo we have included:

  • Two families of the new standard layout. This layout adds a layer of data bits around the outside of the tag border, increasing data density, and the number of possible tags, at the cost of a slight decrease in detection distance.
  • Two families of circular tags.
  • One family which has a hole in the middle. This could be used for example for drone applications by placing different sized tags inside of each other to allow detection over a wide range of distances.

You can generate your own tag families using our other repo, AprilTag-Generation.

Pose Estimation

We have added methods to estimate the 3d pose of the AprilTag given camera parameters and the size of the tag. Sample code is as follows:

// First create an apriltag_detection_info_t struct using your known parameters.
apriltag_detection_info_t info;
info.det = det;
info.tagsize = tagsize;
info.fx = fx;
info.fy = fy; = cx; = cy;

// Then call estimate_tag_pose.
apriltag_pose_t pose;
double err = estimate_tag_pose(&info, &pose);

// Do something with pose.

You can also call estimate_tag_pose_orthogonal_iteration which allows the user to specify the number of iterations used and also returns both possible solutions for the tag pose along with their errors.

Upgrading from AprilTag 3

For most use-cases this should be a drop in replacement.

  • The options refine_decode, refine_pose, and black_border have been removed.
  • If you have generated your own families, you will need to regenerate the c code for those families. The java code however does not need to be regenerated so this should be quick and easy.


The default installation will place headers in /usr/local/include and shared library in /usr/local/lib. It also installs a pkg-config script into /usr/local/lib/pkgconfig. Be aware that there are some larger tag families which may take a long time to build. If you do not want to use these tag families then you can speed up the installation by deleting the files tagCircle49h12.c, tagCircle49h12.h, tagCustom48h12.c, tagCustom48h12.h, tagStandard52h13.c, and tagStandard52h13.h before installing.

$ make
$ sudo make install

To install to a different directory than /usr/local:

$ PREFIX=/some/path sudo make install


A basic AprilTag application can be seen in example/apriltag_demo.c.

Initialization: instantiate a detector and at least one tag family.

apriltag_detector_t *td = apriltag_detector_create();
apriltag_family_t *tf = tag36h11_create();
apriltag_detector_add_family(td, tf);

Some tag detector parameters can be set at this time. The default parameters are the recommended starting point.

td->quad_decimate = 2.0;
td->quad_sigma = 0.0;
td->refine_edges = 1;
td->decode_sharpening = 0.25;

Increase the image decimation if faster processing is required; the trade-off is a slight decrease in detection range. A factor of 1.0 means the full-size input image is used.

Some Gaussian blur (quad_sigma) may help with noisy input images.

Detection: a single one-line call will process an input image and return a list of detections.

zarray_t *detections = apriltag_detector_detect(td, im);

for (int i = 0; i < zarray_size(detections); i++) {
    apriltag_detection_t *det;
    zarray_get(detections, i, &det);

    // Do something with det here


zarray is a container class which is included with apriltag. To process through the list of detections, use zarray_get, as illustrated above.

The caller is responsible for freeing detections by calling apriltag_detections_destroy().

Cleanup: free the detector and tag family when done.


OpenCV Integration

Note that this library has no external dependencies. Most applications will require, at minimum, a method for acquiring images.

See example/ for an example of using AprilTag in C++ with OpenCV. This example application can be built by executing the following:

$ cd examples
$ make opencv_demo

Image data in a cv::Mat object can be passed to AprilTag without creating a deep copy. Simply create an image_u8_t header for the cv::Mat data buffer:

cv::Mat img;

image_u8_t img_header = { .width = img.cols,
    .height = img.rows,
    .stride = img.cols,
    .buf =

Wiki Tutorials

See ROS Wiki Tutorials for more details.

Source Tutorials

Not currently indexed.

Package Dependencies

No dependencies on ROS packages.

System Dependencies


Dependant Packages

Launch files

No launch files found


No message files found.


No service files found


No plugins found.

Recent questions tagged apriltag at Robotics Stack Exchange