visp_auto_tracker wraps model-based trackers provided by ViSP visual servoing library into a ROS package. The tracked object should have a QRcode, Flash code or April tag pattern. Based on the pattern, the object is automaticaly detected. The detection allows then to initialise the model-based trackers. When lost of tracking achieves a new detection is performed that will be used to re-initialize the tracker.
This computer vision algorithm computes the pose (i.e. position and orientation) of an object in an image. It is fast enough to allow object online tracking using a camera.
This package is composed of one node called 'visp_auto_tracker'. The node tries first to detect the QRcode, the Flash or the April tag code associated to the object. Once the detection is performed, the node tracks the object. When a lost of tracking occurs the node tries to detect once again the object and then restart a tracking.
The viewer comming with visp_tracker package can be used to monitor the tracking result.
- [Project webpage on ros.org: tutorial and API reference] ros-homepage
- [Project webpage: source code download, bug report] github-homepage
This package contains submodules. It can be compiled like any other ROS package using catkin_make.
visp_auto_tracker depends on visp_bridge and visp_tracker packages available from https://github.com/lagadic/vision_visp (indigo-devel branches).
visp_auto_tracker depends also on libdmtx-dev and libzbar-dev system dependencies. To install them run:
$ sudo apt-get install libdmtx-dev libzbar-dev
Supposed you have a catkin work space just run:
$ cd ~/catkin_ws/src
$ git clone -b indigo-devel https://github.com/lagadic/vision_visp.git
$ cd ..
$ catkin_make --pkg visp_auto_tracker
The documentation is available on the project [ROS homepage] ros-homepage.
For more information, refer to the [ROS tutorial] ros-tutorial-building-pkg.