All notable changes to this project will be documented in this file. The format is based on Keep a Changelog.
- Add method
get_debug_image
toTriCameraObjectTrackerDriver
and a script that uses it to create debug images from the cameras (this makes diagnosing issues with the object detection easier). - Support
trifinger_cameras::Settings
for changing camera configuration (see documentation oftrifinger_cameras
). - Provide information about frame rate and camera parameters via
get_sensor_info()
like intrifinger_cameras
. For this, overloaded constructors are added to the driver classes which accept camera info files instead of device IDs. - Add a "backend-only" mode to
demo_cameras
. Use this if the front end is running in a separate process.
- The
ProgramOptions
class has been moved to its own package (cli_utils). Thus it is removed from this package and the one from cli_utils is used instead. - Option
--multi-process
fromdemo_cameras
. Use--frontend-only
instead.
- pybind11 build error on Ubuntu 22.04
1.1.0 - 2022-07-29
- Parameter to
CubeVisualizer::draw_cube
to set opacity of the visualisation. CubeVisualizer::draw_circle
to draw a circle at the objects position (e.g. for cases where the orientation does not matter).
1.0.0 - 2022-06-28
- Executable
benchmark_with_logfile
to benchmark the time needed for the object detection using a given camera log file.
- Build models for all supported object types and select the type at run-time (instead of using a build-flag). For this the API of several classes had to be changed as the object model now needs to be passed as argument.
- Update the colour segmentation model for cube v2. The new model is trained on a larger dataset that contains some images of a dirty cube. This should make the model more robust against dirt/dust on the cube.
- Set
filtered_object_pose
in observations when using the PyBullet camera driver.
0.2.0 - 2021.08.04
There is no changelog for this or earlier versions.