if listening to /cob_people_detection/face_recognizer/face_recognitions there is no frame_id set in pose. (Although frame_id is set in general header). --> Please set both

Hello guys!

I'm trying to install cob_people_perception on ROS Lunar.

I have followed the instruction there http://wiki.ros.org/cob_people_detection, beeing aware that those are relevant for ROS Indigo.

Anyway, I get to the following error:

In file included from /home/pierre/catkin_ws/src/cob_people_perception/cob_people_tracking_filter/src/uniform_vector.cpp:37:0:
/home/pierre/catkin_ws/src/cob_people_perception/cob_people_tracking_filter/include/cob_people_tracking_filter/uniform_vector.h:40:21: fatal error: pdf/pdf.h: No such file or directory

I've looked for the file pdf.h and I have it:

$ locate -b '\pdf.h'
/opt/ros/lunar/include/bfl/pdf/pdf.h

I guess /opt/ros/lunar/include is not in the include path.

Any idea?

When first time I installed these cob_people packages, everything worked well. But due to some reason it is not working at all. When I run : roslaunch cob_people_detection people_detection.launch OR roslaunch cob_people_detection people_detection.launch using_nodelets:=true it shows nothing in the display window. The display window is like this:

It shows an error : [ERROR] [1464618029.566038590]: Recognition model not trained

Please note that roslaunch openni_launch openni.launch depth_registration:=true working perfectly and loading drivers successfully. Also kinect is connected.
Also when I write this command: rosrun cob_people_detection people_detection_client then it says No connection to server 'load_model_server'.

I am new. Please help what is wrong. My /home/bacha/catkin_ws/src/cob_people_perception/cob_people_detection/common/files/training_data folder is empty.
Bacha

Folks,

Check in line 192 of cob_people_perception/cob_openni2_tracker/ros/src/body_tracker.cpp where the buffer size is Zero (infinite size). It should be at least 1, as this can cause issue when there is a lot of subscribers for this topic.

I asked this question on answers.ros.org but I'm not sure the COB developers monitor that forum so I repeat it here.

I am using the most excellent cob_people_detection package under ROS Indigo (Ubuntu 14.04). The package does a nice job of detecting heads and faces when using my robot's Xtion Pro Live camera. However, it seems that the head and face detections are published only as 2D images or Rects and the distance to the head or face is not published. Since depth is used as part of the head detection algorithm, I'm wondering why the distance to the detected head is not published. Or am I missing something?

Ideally, it would be nice to get the position of the tracked head published as a PoseStamped message so it could be placed relative to the robot within a mapped environment.

Thanks!

Hi,
I am so interested in your project . I have a code for activity recognition using 3d body joints published by tf transformer (using openni tracker) .My goal is to try to synchronize the output face and head position of your project with the skeleton extracted in my project . Can You help me please . Sorry for disturbing and thank you for advance .

I had a few questions on executing cob_people_detection. I was easily able to follow the wiki ros tutorial and get the face detection and recognition working. However, I could not figure out how to do tracking? Could you help me with it.

I also tried to run the leg_detector individually and nothing really happened. I followed the following process to run the leg detector,

Terminal 1: roscore
Teminal 2: roslaunch openni_launch openni.launch
terminal 3: roslaunch cob_leg_detector leg_detector.launch.

And nothing really happened. Am I supposed to visualize it in a different way?

And my last question is how do I get the 3D coordinates of the face detected?

I would be really grateful if you could help me out. I am new to ROS and am not able to completely wrap my head around it.

Thank you,
Nisha

I ran the program following the steps in the ROS wiki. I ran it once successfully and it crashed as soon as I loaded the recognition model I just built by taking automatic pictures. Now every time I try to run rosrun cob_people_detection people_detection_client, I get the error message No connection to server 'load_model_server'..

I think I could correct the issue if you informed me where this model that it reports is stored. I tried looking through the source code and narrowed it down to loadActionModel.h, but couldn't find anywhere in the code where a model was being loaded. I would then make a pull request to clarify the error message.

@ipa-fmw @ipa-rmb
In order to be able to switch back to indigo_dev for MSH (and then removing msh_dev), there is only little changes needed...

1. According to https://github.com/ipa320/cob_people_perception/compare/indigo_dev...msh_dev?expand=1, the only relevant change in msh_dev is to disable face_recognition. We could fix this by using our msh-specific yaml file

2. According to https://github.com/ipa320/cob_people_perception/compare/msh_dev...indigo_dev?expand=1, the only relevant change in indigo_dev that affects the MSH usecase is the modified topic namespaces. We could fix this by either making the topics non-global, so that the namespace could be modified or we adjust the namespaces in msh application

in #63 Travis CI tests showed that this branch needs to properly be migrated to kinetic/xenial on a new kinetic_dev branch

Does anybody knows what this comes out "ADD_LIBRARY for library..."?

-- +++ processing catkin metapackage: 'cob_people_perception'
-- ==> add_subdirectory(cob_people_perception/cob_people_perception)
-- +++ processing catkin metapackage: 'cob_perception_common'
-- ==> add_subdirectory(cob_perception_common/cob_perception_common)
-- +++ processing catkin package: 'libnite2'
-- ==> add_subdirectory(cob_people_perception/libnite2)
You have called ADD_LIBRARY for library NiTE2 without any source files. This typically indicates a problem with your CMakeLists.txt file

I've exported the ROS_MASTER_URI on both computers and the ROS_HOSTNAME on each of them. Everything seems to be running fine with no errors. I have access on the depth and rgb images sent by the core machine, plus on the sensor_message_gateway image outputs. Still, nothing shows up on the image_view when running cob_people_detection.
I am running the astra camera (astra.launch) on the core machine.
And the people detection (people_detection.launch) on the second one.

I have followed the "Quick Start" steps on the wiki. I launch the people_detection.launch file and the display window opens fine and displays the cameras view. However I get stuck on the step to activate sensor message gateway.
I do not get "Gateway successfully opened" message after I have entered the target rate frame (see image attached).

Additionally, when I Ctrl+C from the terminal running "people_detection.launch" I get the message "couldn't set parameters at node sensor_message_gateway/sensor_message_gateway"

Any help getting this solved is appreciated

• Roberto

Hi,

I have a project for activities analysis and face recognition . I just link my project of activity recognition which is subscribed to tf topic and the project of "cob_people_perception" but the problem is there no synchronization to link every skeleton with it's appropriate face. I will be so grateful if you help me how to synchronize tf topic with cob_perception_msgs::ColorDepthImageArray topic with tf:MessageFilter . Bellow is the code where I want to make synchronization .Thank you for advance . I try with this code in the file face_recognizer_node.cpp in main function but it shows unkown error .
face_recognizer_node.cpp.zip

It is possible that the Asus Xtion and the Kinect sensor and maybe even the openni_launch and openni2_launch drivers output data with different definitions of coordinate systems (or at least in their internal representations). If that is true, the cob_openni2_tracker software needs to identify the utilized device or driver and convert coordinates accordingly. The current setting is good for Asus Xtion with openni2_launch. There is no Kinect available for testing at this moment.

In your wiki post, you say that to run the program, you should follow these steps:
roslaunch openni_launch openni.launch
roslaunch cob_people_detection people_detection.launch
rosrun cob_people_detection people_detection_client

However, in the wiki, you cite these steps instead:
roslaunch cob_people_detection openni.launch
roslaunch cob_people_detection people_detection.launch
rosrun cob_people_detection people_detection_client

Could you please clarify which one is correct? Both are giving me problems.

Whenever I run this package with ROS, I notice a surprisingly high CPU usage, regardless of the frequency of the message gateway that I select. For example, at 5 Hz, I see a CPU usage of around 85% for my 4 cores of my Intel® Core™2 Quad CPU Q9650 @ 3.00GHz. I have a GeForce GTX 285 as my GPU, so I'm assuming this would take most of the load off the processing anyways, but I can't confirm for sure since there is no way for me to monitor my GPU usage.

Is this expected or is this an indication that there is something wrong with my installation? Alternatively, if it is expected, is there anything I can help out that is being investigated that might bring the usage down?

Folks,

I am running the cob_oppenni2_tracker, but it crashes after a certain amount of time. I pretty much the installation instructions with not modifications: http://wiki.ros.org/cob_people_detection.

I am using an Xtion sensor (the 640 x 480 resolution one) connected to USB 2.0. Any compatibility (or resource sharing) issues? Is it maybe an issue with running openni2_launch alongside cob_openni2_tracker nodes? A "Timeout" issues if a user is not found, perhaps?

This is the my entire error log I get (sorry if it's too long):

[FATAL] [1468014962.668436134]: Failed to load nodelet '/camera/depth_points of type depth_image_proc/point_cloud_xyz to manager camera_nodelet_manager' [FATAL] [1468014962.668452424]: Failed to load nodelet '/camera/rectify_color of type image_proc/rectify to manager camera_nodelet_manager' [FATAL] [1468014962.668610743]: Failed to load nodelet '/camera/depth_registered_rectify_depth of type image_proc/rectify to manager camera_nodelet_manager' [FATAL] [1468014962.670175538]: Failed to load nodelet '/camera/depth_registered_metric of type depth_image_proc/convert_metric to manager camera_nodelet_manager' [FATAL] [1468014962.670253997]: Failed to load nodelet '/body_tracker of type cob_openni2_tracker/body_tracker_nodelet to manager /camera/camera_nodelet_manager' [FATAL] [1468014962.670262776]: Failed to load nodelet '/camera/points_xyzrgb_hw_registered of type depth_image_proc/point_cloud_xyzrgb to manager camera_nodelet_manager' [FATAL] [1468014962.670381691]: Failed to load nodelet '/camera/depth_metric_rect of type depth_image_proc/convert_metric to manager camera_nodelet_manager' [camera/camera_nodelet_manager-2] process has died [pid 29688, exit code -11, cmd /opt/ros/indigo/lib/nodelet/nodelet manager __name:=camera_nodelet_manager __log:=/home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-camera_nodelet_manager-2.log]. log file: /home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-camera_nodelet_manager-2*.log [camera/rectify_color-4] process has died [pid 29690, exit code 255, cmd /opt/ros/indigo/lib/nodelet/nodelet load image_proc/rectify camera_nodelet_manager --no-bond image_mono:=rgb/image_raw image_rect:=rgb/image_rect_color __name:=rectify_color __log:=/home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-rectify_color-4.log]. log file: /home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-rectify_color-4*.log [camera/depth_registered_metric-12] process has died [pid 29755, exit code 255, cmd /opt/ros/indigo/lib/nodelet/nodelet load depth_image_proc/convert_metric camera_nodelet_manager --no-bond image_raw:=depth_registered/image_raw image:=depth_registered/image __name:=depth_registered_metric __log:=/home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-depth_registered_metric-12.log]. log file: /home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-depth_registered_metric-12*.log [body_tracker-17] process has died [pid 29810, exit code 255, cmd /opt/ros/indigo/lib/nodelet/nodelet load cob_openni2_tracker/body_tracker_nodelet /camera/camera_nodelet_manager __name:=body_tracker __log:=/home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/body_tracker-17.log]. log file: /home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/body_tracker-17*.log [camera/depth_metric_rect-6] process has died [pid 29698, exit code 255, cmd /opt/ros/indigo/lib/nodelet/nodelet load depth_image_proc/convert_metric camera_nodelet_manager --no-bond image_raw:=depth/image_rect_raw image:=depth/image_rect __name:=depth_metric_rect __log:=/home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-depth_metric_rect-6.log]. log file: /home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-depth_metric_rect-6*.log [camera/depth_points-8] process has died [pid 29721, exit code 255, cmd /opt/ros/indigo/lib/nodelet/nodelet load depth_image_proc/point_cloud_xyz camera_nodelet_manager --no-bond image_rect:=depth/image_rect_raw points:=depth/points __name:=depth_points __log:=/home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-depth_points-8.log]. log file: /home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-depth_points-8*.log [camera/depth_registered_rectify_depth-9] process has died [pid 29732, exit code 255, cmd /opt/ros/indigo/lib/nodelet/nodelet load image_proc/rectify camera_nodelet_manager --no-bond image_mono:=depth_registered/image_raw image_rect:=depth_registered/hw_registered/image_rect_raw __name:=depth_registered_rectify_depth __log:=/home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-depth_registered_rectify_depth-9.log]. log file: /home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-depth_registered_rectify_depth-9*.log [camera/points_xyzrgb_hw_registered-10] process has died [pid 29742, exit code 255, cmd /opt/ros/indigo/lib/nodelet/nodelet load depth_image_proc/point_cloud_xyzrgb camera_nodelet_manager --no-bond rgb/image_rect_color:=rgb/image_rect_color rgb/camera_info:=rgb/camera_info depth_registered/image_rect:=depth_registered/hw_registered/image_rect_raw depth_registered/points:=depth_registered/points __name:=points_xyzrgb_hw_registered __log:=/home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-points_xyzrgb_hw_registered-10.log]. log file: /home/luis/.ros/log/1630b54a-452d-11e6-8d96-90004e042890/camera-points_xyzrgb_hw_registered-10*.log

Hi there.
Previously I used the cob_people_perception package successfully. However, I formatted my computer and had to reinstall it. Since then I'm getting the following error:

/ opt / ros / kinetic / lib / nodelet / nodelet: symbol lookup error: /home/andres/catkin_ws/devel/lib//libbody_tracker_nodelet.so: undefined symbol: _ZN2cv7waitKeyEi [kinect / kinect_nodelet_manager-2] process has died [pid 13262, exit code 127, cmd / opt / ros / kinetic / lib / nodelet / nodelet manager

Help would be great!
Thanks

Hi,
Thank you so your project and for your help. But there is just a small problem . While recognition , the algorithm detect unknown people as one between saved people and there is some confusion in recognition of the true person Can you please help me where can I improve the accuracy of recognition ? thank your for advance

Hi, thank you for your useful package.
Although I really hope to use your program in my project, I encounter some problem.

[My computer environment]
OS : Ubuntu 14.04
ROS: Indigo
Camera: Kinect XBOX360

reference link: http://wiki.ros.org/cob_people_detection

[Problem issue]
When launching this command ~$ roslaunch cob_people_detection people_detection.launch ,
these Errors occur.

and I can't see any image from an pop-up window

What factors do make these problem??

Folks,

Why can't I see the joints being published? It's tracking the user, but I cannot see anything in the "rostopic echo tf..."

When I do "rosrun rqt_tf_tree rqt_tf_tree" I only see the camera links. Why not the transforms?

I made the ros_package "cob_people_detection" in groovy and I got an error.

[rospack] Error: package/stack 'cob_people_detection' depends on non-existent package 'kdl' and rosdep claims that it is not a system dependency. Check the ROS_PACKAGE_PATH or try calling 'rosdep update'

CMake Error at /opt/ros/groovy/share/ros/core/rosbuild/public.cmake:129 (message):

Failed to invoke rospack to get compile flags for package
'cob_people_detection'. Look above for errors from rospack itself.
Aborting. Please fix the broken dependency!

Call Stack (most recent call first):
/opt/ros/groovy/share/ros/core/rosbuild/public.cmake:203 (rosbuild_invoke_rospack)
CMakeLists.txt:16 (rosbuild_init)

I have already installed the ros-groovy-kdl, but the error still exist. WHY?

Originally issued by RoboEvangelist in the wrong repository, see here ipa320/cob_perception_common#63 .

Folks,

I installed the package, but it overridden my NiTE1.x installations, and also changed the paths in my cmake_prefix_path environment variable. Now I cannot run my other packages, even after uninstalling the cob_* package.
I wonder if there is a way to install the NiTE2 from this repository in a local way so that it does not affect the rest of the package in ROS? By this I mean no Make/CMakeList hacking required from the user point of view. Or at least give the option to the user to install it locally, or globally. I don't think it's that difficult to do that.

Other than that, the face detection is pretty cool!

Thanks,

Hi

I've been trying to compile the cob_people_perception code but I keep running into errors:

lex@lex:~/Downloads$ cd ~/catkin_ws/ && catkin_make -DCMAKE_BUILD_TYPE="Release"
Base path: /home/lex/catkin_ws
Source space: /home/lex/catkin_ws/src
Build space: /home/lex/catkin_ws/build
Devel space: /home/lex/catkin_ws/devel
Install space: /home/lex/catkin_ws/install

Running command: "cmake /home/lex/catkin_ws/src -DCMAKE_BUILD_TYPE=Release -DCATKIN_DEVEL_PREFIX=/home/lex/catkin_ws/devel -DCMAKE_INSTALL_PREFIX=/home/lex/catkin_ws/install -G Unix Makefiles" in "/home/lex/catkin_ws/build"

-- Using CATKIN_DEVEL_PREFIX: /home/lex/catkin_ws/devel
-- Using CMAKE_PREFIX_PATH: /home/lex/catkin_ws/src/cob_people_perception
-- Using PYTHON_EXECUTABLE: /usr/bin/python
-- Using Debian Python package layout
-- Using empy: /usr/bin/empy
-- Using CATKIN_ENABLE_TESTING: ON
-- Call enable_testing()
-- Using CATKIN_TEST_RESULTS_DIR: /home/lex/catkin_ws/build/test_results
-- Found gtest sources under '/usr/src/gtest': gtests will be built
-- Using Python nosetests: /usr/bin/nosetests-2.7
-- catkin 0.6.14
-- BUILD_SHARED_LIBS is on
-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- ~~ traversing 13 packages in topological order:
-- ~~ - cob_people_detection
-- ~~ - cob_people_perception (metapackage)
-- ~~ - cob_people_tracking_filter
-- ~~ - cob_leg_detection
-- ~~ - face_recognition
-- ~~ - iai_kinect2 (metapackage)
-- ~~ - kinect2_registration
-- ~~ - kinect2_bridge
-- ~~ - kinect2_calibration
-- ~~ - kinect2_viewer
-- ~~ - libnite2
-- ~~ - cob_openni2_tracker
-- ~~ - usb_cam
-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- +++ processing catkin package: 'cob_people_detection'
-- ==> add_subdirectory(cob_people_perception/cob_people_detection)
-- Using these message generators:
-- Boost version: 1.54.0
-- Found the following Boost libraries:
-- signals
-- system
-- filesystem
-- Generating .msg files for action cob_people_detection/addData /home/lex/catkin_ws/src/cob_people_perception/cob_people_detection/action/addData.action
-- Generating .msg files for action cob_people_detection/deleteData /home/lex/catkin_ws/src/cob_people_perception/cob_people_detection/action/deleteData.action
-- Generating .msg files for action cob_people_detection/getDetections /home/lex/catkin_ws/src/cob_people_perception/cob_people_detection/action/getDetections.action
-- Generating .msg files for action cob_people_detection/loadModel /home/lex/catkin_ws/src/cob_people_perception/cob_people_detection/action/loadModel.action
-- Generating .msg files for action cob_people_detection/updateData /home/lex/catkin_ws/src/cob_people_perception/cob_people_detection/action/updateData.action
CMake Error at /opt/ros/indigo/share/genmsg/cmake/genmsg-extras.cmake:268 (message):
Could not find 'share/std_msgs/cmake/std_msgs-msg-paths.cmake' (searched in
'/home/lex/catkin_ws/devel').
Call Stack (most recent call first):
cob_people_perception/cob_people_detection/CMakeLists.txt:87 (generate_messages)

-- Configuring incomplete, errors occurred!
See also "/home/lex/catkin_ws/build/CMakeFiles/CMakeOutput.log".
See also "/home/lex/catkin_ws/build/CMakeFiles/CMakeError.log".
Invoking "cmake" failed

Could you help me look into the error log please?
Thank you!

Hi,
I amso intersted in your project . But I have a poblem , I will so grateful i you help me . firstely I run
roslaunch openni_launch openni.launch device_id:=#2 an it is ok . Then I run "roslaunch cob_people_detection people_detection.launch" and I I have a problem .
System : elemeentary OS
Version Ros : kinetic
Thank you for advance

Due to the lack of a README, this package is a little confusing to install. Based on this wiki, I know I have to clone your repository and build the package.

I've cloned the repository using git clone https://github.com/ipa320/cob_people_perception.git.
I then created a directory called build cd cob_people_perception && mkdir build && cd build
I then started Cmake cmake .. and then sudo make .. and desperately sudo make install, but launching the service (after launching openni of course) with roslaunch cob_people_detection people_detection.launch failed.

It then tried again without making the build directory cmake . and sudo make and desperately sudo make install, but launching the service still failed.

Judging by the output of make -d there doesn't seem to be any obvious errors.

By failed, I mean that got the error message [people_detection.launch] is neither a launch file in package [cob_people_detection] nor is [cob_people_detection] a launch file name, even after attempting rosintall cob_people_detection /opt/ros/groovy.

If you could make a quick plain text README, it would be greatly appreciated. I would even submit a pull request to format it and expand on it!