hri_rviz

This package provides a list of rviz2 plugins for human-related data visualisation. It is part of the ROS4HRI ecosystem.

Plugins

Humans

A plugin for visualising 2D information overlayed on a camera stream (ideally, the stream used to detect it). Currently, the plugin can visualise:

• Face bounding boxes
• Face landmarks
• Body bounding boxes
• 2D skeleton keypoints

How to use it

1. If not already available, add a Displays panel;
2. press the Add button at the bottom to istantiate a new plugin;
3. select By topic;
4. among the available topics, select the Humans plugin for the camera stream you are interested in;
5. once created the plugin, select which type of information you want to visualise (face bounging boxes, skeleton landmarks, etc.);
6. enjoy!

Skeletons3D

A plugin for visualising the estimated 3D poses of the detected humans.

How to use it

1. If not already available, add a Displays panel;
2. press the Add button at the bottom to istantiate a new plugin;
3. select By display type;
4. select Skeletons3D;
5. enjoy!

TF_HRI

A plugin for visualising the human-related TF frames. These are higly dynamical, appearing and disappearing in a matter of seconds. Using the classic TF plugin would result in a crowded and chaotic frames visualisation. This plugin:

• looks over the detected faces and bodies;
• only displays the face and bodies TF frames for the currently detected bodies and faces;
• readily remove the TF frames for those bodies and faces that are no more tracked, avoiding the disappearing phase observed in the original TF frame for the non-updated frames. It is possible to select which human frames to visualise among:
• face frames;
• gaze frames;
• body frames;

How to use it

1. If not already available, add a Displays panel;
2. press the Add button at the bottom to istantiate a new plugin;
3. select By display type;
4. select TF_HRI;
5. enjoy!

Test

To test the hri_rviz plugins:

1. Download hri_face_detect and hri_fullbody;
2. build them;
3. start an RGB camera stream;
4. start face and body detection:
   • ros2 launch hri_face_detect face_detect.launch.py filtering_frame:=<camera_frame> rgb_camera:=<rgb_camera_stream_ns>
   • ros2 launch hri_fullbody hri_fullbody.launch.py rgb_camera:=<rgb_camera_stream_ns> where <rgb_camera_stream_ns> is the RGB camera stream namespace (e.g., /camera/color). This expects the raw RGB images to be published on <rgb_camera_stream_ns>/image_raw. Check the launch files parameters for different options.
5. start the plugins as previously described.

Changelog for package hri_rviz

2.1.0 (2024-09-19)

• display expression when available
• Contributors: saracooper

2.0.0 (2024-05-21)

• re-import CHANGELOG from ROS1
• general cleanup linting code + fixing wrong dependencies in package.xml
• reading body description from the new dedicated topic
• add LICENSE and CONTRIBUTING.md
• using rviz-ogre-vendor instead of native ogre libraries see https://github.com/ros2/rviz/issues/876
• documentation
• fixed icons
• added copyright
• TF (HRI) porting
• Skeletons3D porting
• Humans porting
• Contributors: Séverin Lemaignan, lorenzoferrini

0.4.2 (2023-10-18)

• moving from non-normalized to normalized facial landmarks
• Contributors: lorenzoferrini

0.4.1 (2023-07-05)

• changed RoI type to normalized one
• Contributors: Luka Juricic

0.4.0 (2023-01-18)

• Merge branch 'skeletons' into 'main' skeleton extension for Humans plugin See merge request ros4hri/hri_rviz!5
• joint visualization as circles
• Optimized iteration over bodies for skeleton representation
• Introducing skeleton visualization It is now possible to visualize 2D skeletons using the humans rviz plugin.
• add LICENSE
• Contributors: Séverin Lemaignan, lorenzoferrini

0.3.1 (2022-07-12)

• set license to BSD
• Contributors: Séverin Lemaignan

0.3.0 (2022-07-12)

• Added PAL Robotics copyright
• Refactored skeleton frames testing The new version takes a frame name, verifies if it has more than the minimum number of characters for a skeleton frame, checks if the name without the last 5 characters belongs to the skeleton ROS4HRI naming convention and if it belongs to a currently tracked body.
• bodies -> skeletons Bodies frames are now referenced as skeleton frame, for clarity.
• Redefined class name for the plugin To avoid confusion with the already existing TF plugin, the plugin will now be visualized as TF (HRI) in rviz
• First complete hri_tf version
• Fixed skeletons appearing when plugin was disabled
• Contributors: Séverin Lemaignan, lorenzoferrini

0.2.0 (2022-03-06)

• Facial landmarks visualization
• raw pointers to std::share_ptr
• add basic README
• Contributors: Séverin Lemaignan, lorenzoferrini

0.1.3 (2022-01-21)

• update to libhri 0.2.3
• Plugin renaming: Faces --> Humans
• Contributors: Séverin Lemaignan, lorenzoferrini

0.1.2 (2022-01-14)

• replace hri_msgs::RegionOfInterestStamped by sensor_msgs::RegionOfInterest Follows changes in hri_msgs 0.2.0
• Contributors: Séverin Lemaignan

0.1.1 (2022-01-13)

• remove code that relies on too-recent rviz APIs, to ensure compat with melodic/noetic
• code formatting + PAL copyright
• add missing dependencies + fix flags
• Contributors: Séverin Lemaignan

0.1.0 (2022-01-13)

• port the plugin to libhri; use colors generated from ID for bbs
• Multi-Skeleton 3D visualization plugin
• Faces Plugin as ImageDisplay-like object Faces Plugin displaying multiple faces simultaneously, with random colors for each bounding

File truncated at 100 lines see the full file