wg-perception / capture Goto Github PK

HI

I am trying to install ork from this link

I did the following and every thing looks fine

mkdir ws && cd ws
wstool init src https://raw.github.com/wg-perception/object_recognition_core/master/doc/source/ork.rosinstall.kinetic.plus
cd src && wstool update -j8
cd .. && rosdep install --from-paths src -i -y
catkin_make
source devel/setup.bash
sudo pip install -U breathe catkin-sphinx sphinxcontrib-programoutput
cd build
make doxygen

BUT the last step creates error as follow

Any help please ??!!

abdu@grasp-uoit:~/ws/build$ make sphinx-doc
Traceback (most recent call last):
  File "/usr/local/bin/sphinx-build", line 11, in <module>
    load_entry_point('Sphinx==1.7.5', 'console_scripts', 'sphinx-build')()
  File "/usr/local/lib/python2.7/dist-packages/Sphinx-1.7.5-py2.7.egg/sphinx/cmd/build.py", line 38, in main
    return build_main(argv)
  File "/usr/local/lib/python2.7/dist-packages/Sphinx-1.7.5-py2.7.egg/sphinx/cmd/build.py", line 22, in build_main
    from sphinx import cmdline
  File "/usr/local/lib/python2.7/dist-packages/Sphinx-1.7.5-py2.7.egg/sphinx/cmdline.py", line 19, in <module>
    from docutils.utils import SystemMessage
  File "/home/abdu/py3-gpu/lib/python3.5/site-packages/docutils/utils/__init__.py", line 21, in <module>
    import docutils.io
  File "/home/abdu/py3-gpu/lib/python3.5/site-packages/docutils/io.py", line 349
    (self.destination.mode, mode)), file=self._stderr)
                                        ^
SyntaxError: invalid syntax
ork_tabletop/CMakeFiles/object_recognition_tabletop-sphinx.dir/build.make:57: recipe for target 'object_recognition_tabletop-sphinx' failed
make[3]: *** [object_recognition_tabletop-sphinx] Error 1
CMakeFiles/Makefile2:11623: recipe for target 'ork_tabletop/CMakeFiles/object_recognition_tabletop-sphinx.dir/all' failed
make[2]: *** [ork_tabletop/CMakeFiles/object_recognition_tabletop-sphinx.dir/all] Error 2
CMakeFiles/Makefile2:530: recipe for target 'ecto/CMakeFiles/sphinx-doc.dir/rule' failed
make[1]: *** [ecto/CMakeFiles/sphinx-doc.dir/rule] Error 2
Makefile:316: recipe for target 'sphinx-doc' failed
make: *** [sphinx-doc] Error 2

When running
rosrun object_recognition_capture capture -i template -o stuff.bag the script crashes with

Satisfied, total observations: 36
terminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<boost::lock_error> >'
  what():  boost::lock_error
Aborted (core dumped)

However, the bag file is produced anyway :)

edit: I just noticed this only happens some times ...

Capture app is crashing with Segmentation fault message. Using gdb we realised that the problem is in file .../cells/PlaneFilter.cpp:148 (process function) when code:

if(*mask != 255)

should be executed. When I try to do

print *mask

in gdb I got "Cannot access memory at address 0x....."

This i get from both, this repository and ros official repository, versions (both are the newest version).

I'm trying to follow instructions at

http://ecto.willowgarage.com/recognition/ros_quickguide.html

taking into account considerations made at

http://tinyurl.com/9tk9wuk

• Installing ros precise packages

  • orb_template-py returns "no madule named roscompat"
  • in your last commits, the imports roscompat is removed, so removing that results in a "no module named ecto_openni"
  • compiling and installing by hand ecto_openni does not help, and a new arror raises

• Installing sources

• catkin package resides in a different repo. I install from github.com/ros/catkin.git

• cannot run the cmake..

  include could not find load file:

  /home/galenya/iri/code/object_recognition/build/buildspace/share/ecto/cmake/rosbuild_lite.cmake
  Call Stack (most recent call first):
  build/buildspace/share/ecto/cmake/ectoConfig.cmake:133 (include)
  ecto_opencv/CMakeLists.txt:33 (find_package)

that corresponds to a

find_package(ecto REQUIRED)

instruction.

Could you help me indicting which version of catkin/catkin_pkg/ecto should I use and how I've to install them?

Thanks for your help.

Can we use the ZED camera for doing this?

I am not sure if it is possible to "close the loop" in creating the 3D object model. If you put the object on the capture board and scan it from all sides you will not capture any data on the bottom of the object. Is there a way to get around this issue?

Hello,

I am trying to run the ORK package as it is said in the tutorials and using the fiducial grid that is proposed, but when I try to capture the object and create the bag of the object I have the following error:

rosrun object_recognition_capture capture --seg_z_min 0.01 -o silk.bag
Registration? 1
Sync? 0

1. device on bus: 1 @ 29 with serial number: A00364A04947053A Microsoft : Xbox NUI Camera
   Time diff max: 16.8667
   Registered:on Supported: 1
   Setting registration on
   Setting sync off
   Novel view, total:1
   Opening bag: silk.bag
   Novel view, total:2
   Novel view, total:3
   Novel view, total:4
   Novel view, total:5
   Novel view, total:6
   Novel view, total:7
   Novel view, total:8
   Novel view, total:9
   Novel view, total:10
   Novel view, total:11
   Novel view, total:12
   Novel view, total:13
   Novel view, total:14
   Novel view, total:15
   Novel view, total:16
   Novel view, total:17
   Novel view, total:18
   Novel view, total:19
   Novel view, total:20
   Novel view, total:21
   Novel view, total:22
   Novel view, total:23
   Novel view, total:24
   Novel view, total:25
   Novel view, total:26
   Novel view, total:27
   Novel view, total:28
   Novel view, total:29
   Novel view, total:30
   Novel view, total:31
   Novel view, total:32
   Novel view, total:33
   Novel view, total:34
   Novel view, total:35
   Novel view, total:36
   Satisfied, total observations: 36
   QObject::~QObject: Timers cannot be stopped from another thread
   Segmentation fault (core dumped)

So, apparently the images are taken but there is some kind of error maybe in the installation of the package. Then, the same error occurs when i try to create the mesh. Does anyone know what could it be?

I did tabletop with coke detection, everything is fine. Now, I am trying to run object_recognition capture following this tutorial

but when I run this

rosrun object_recognition_capture orb_template -o my_textured_plane

I get the following error, the reason is the camera is not detected, I am using Kinect v2

Any help ??

abdu@grasp-uoit:~$ rosrun object_recognition_capture orb_template -o my_textured_plane
Listening for key: s on imshow:save
Registration? 1
Sync? 0
Traceback (most recent call last):
  File "/home/abdu/ws/src/ork_capture/apps/orb_template", line 108, in <module>
    run_plasm(options, plasm, locals=vars())
  File "/home/abdu/ws/src/ecto/python/ecto/opts.py", line 85, in run_plasm
    sched.execute(options.niter)
ecto.CellException:            exception_type  CellException
[cell_name] = Source

[cell_type] = ecto::py::BlackBox

[function_name] = process_with_only_these_inputs

[type] = std::runtime_error

[what] =            exception_type  CellException
[cell_name] = <class 'ecto_openni.OpenNICapture'>

[cell_type] = ecto_openni::OpenNICapture

[function_name] = configure

[type] = std::runtime_error

[what] = Index out of range.0 devices found.

Hi again,

I tried to run the object_recognition_capture with a template but the mask image remains dark and no capturing is performed:

rosrun object_recognition_capture capture -i my_textured_plane -o orc_scan_dataIntensiv.bag -n 12 --preview --seg_z_min 0.0001

The pose estimation seems to work, the coordinate origin is reprojected correctly onto the template plane:

I tried different optional parameter but nothing changed. I compiled ecto and wg_perception completely from source within a catkin groovy workspace on Ubuntu 12.4, the data comes from a Kinect.

When running

rosrun object_recognition_capture capture --seg_z_min 0.01 -o silk.bag --res sxga --fps 15

The result is

Registration? 1
Sync? 0

1. device on bus: 2 @ 6 with serial number: A00362902372052A Microsoft : Xbox NUI Camera
   Time diff max: 33.5333
   Registered:on Supported: 1
   Setting registration on
   Setting sync off
   OpenCV Error: Sizes of input arguments do not match (The operation is neither 'array op array' (where arrays have the same size and type), nor 'array op scalar', nor 'scalar op array') in binary_op, file /tmp/buildd/ros-fuerte-opencv2-2.4.2-1precise-20130312-1308/modules/core/src/arithm.cpp, line 1021
   Traceback (most recent call last):
   File "/opt/ros/fuerte/share/object_recognition_capture/bin/capture", line 76, in
   run_plasm(options, plasm, locals=dict(plasm=plasm, segmentation=segmentation))
   File "/opt/ros/fuerte/lib/python2.7/dist-packages/ecto/opts.py", line 85, in run_plasm
   sched.execute(options.niter)
   ecto.CellException: exception_type CellException
   cell_name imgproc::BitwiseAnd
   cell_type imgproc::BitwiseAndnks
   function_name process
   type cv::Exception
   what /tmp/buildd/ros-fuerte-opencv2-2.4.2-1precise-20130312-1308/modules/core/src/arithm.cpp:1021: error: (-209) The operation is neither 'array op array' (where arrays have the same size and type), nor 'array op scalar', nor 'scalar op array' in function binary_op

I've no problem running with VGA and 30 fps
btw, my system is ubuntu 12.04 running with ros-fuerte
the package is install from ROS repositories

Any idea? Thanks

Hello,
I am following this tutorial: http://wg-perception.github.io/capture/index.html#ork-capture
When I try to run this command:
rosrun object_recognition_capture capture --seg_z_min 0.01 -o pringles.bag --preview

I get the following error:
Registration? 1 Sync? 0 python: /usr/include/boost/smart_ptr/shared_ptr.hpp:646: typename boost::detail::sp_dereference<T>::type boost::shared_ptr<T>::operator*() const [with T = xn::NodeInfo; typename boost::detail::sp_dereference<T>::type = xn::NodeInfo&]: Assertionpx != 0' failed.
Aborted (core dumped)
`
Do you have any suggestions? Thank you for your help.

I am using Ubuntu 14.04, Ros Indigo, OpenCV 2.4.11 and Python 2.7.6

Details can be found here:
http://answers.ros.org/question/78086/object-recognition-capture-unable-to-find-pose/

I installed the package from ROS official repository.(ROS groovy, Ubuntu 12.04)

ros-groovy-object-recognition-capture 0.2.22-0precise-20130727-2206-+0000
ros-groovy-object-recognition-core 0.5.4-0precise-20130727-2142-+0000
ros-groovy-ecto 0.5.5-0precise-20130721-1603-+0000
ros-groovy-ecto-opencv 0.4.25-0precise-20130727-2052-+0000
ros-groovy-ecto-openni 0.3.9-0precise-20130721-1714-+0000
ros-groovy-ecto-image-pipeline 0.4.12-0precise-20130727-2116-+0000

I tried to run the object_recognition_capture with the dot pattern. I did not print it to its original size but just chose A3 with automatic scale. When running rosrun object_recognition_capture capture -o orc_scan.bag the pattern is not completely recognized:

However, two dots, one from the dark and one from the black side, are reprojected.

The wrong pattern size shouldn't affect the pattern recognition, should it? I guess it would just calculate a wrong distance of the pattern.
Anyway, I tried to change the hard coded pattern parameter in fiducial_pose_est.py by multiplying every number concerned with the pattern by 0.5, which didn't improve the recognition.

I have a white light scanner that combined with its software suite produces high resolution color models.
It would be nice to be able to capture a bag from a model so that training can be done with the high resolution models I have scanned. This seems like it would be a combination of the functionality of ork_renderer and capture.
My interest is to use this for the TOD, although it may also be useful for other detectors.
Has any work been done on this? I would be interested in adding this myself with a little direction.
I apologize if this is the wrong place to post this request.

Hi! I'm using a Realsense D435 Depth Camera, in order to launch the capture process i run
roslaunch realsense2_camera rs_camera.launch
but when i tried to run the capture process by
'rosrun object_recognition_capture orb_template -o my_textured_plane'
it gives me an 0 device found

Listening for key: s on imshow:save
Registration? 1
Sync? 0
Traceback (most recent call last):
  File "/home/xxwang/ork/src/capture/apps/orb_template", line 108, in <module>
    run_plasm(options, plasm, locals=vars())
  File "/opt/ros/kinetic/lib/python2.7/dist-packages/ecto/opts.py", line 85, in run_plasm
    sched.execute(options.niter)
ecto.CellException:            exception_type  CellException
[cell_name] = Source

[cell_type] = ecto::py::BlackBox

[function_name] = process_with_only_these_inputs

[type] = std::runtime_error

[what] =            exception_type  CellException
[cell_name] = <class 'ecto_openni.OpenNICapture'>

[cell_type] = ecto_openni::OpenNICapture

[function_name] = configure

[type] = std::runtime_error

[what] = Index out of range.0 devices found.

I've noticed that in issue #20 they gave a possible method to use ros topics instead of openni, though i can not understand exactly how to do that. I still want to know if it is possible to run capture with realsense camera and realsense-ros package. Thank you and forgive my poor English

When I (according to Quick guide) try Capture a workspace in 'depth' and 'ORB' windows aren't images from kinect. 'depth' window is black, and 'ORB' window is red with resolution and frequency label. But when I press 's' data are saved, and .png image is viewable and correct.

Capture fail to start with:

python: /usr/include/eigen3/Eigen/src/Core/DenseStorage.h:69: Eigen::internal::plain_array<T, Size, MatrixOrArrayOptions, 16>::plain_array() [with T = double, int Size = 16, int MatrixOrArrayOptions = 0]: Assertion `(reinterpret_cast<size_t>(array) & 0xf) == 0 && "this assertion is explained here: " "http://eigen.tuxfamily.org/dox-devel/TopicUnalignedArrayAssert.html" " **** READ THIS WEB PAGE !!! ****"' failed.

Is it a bug in ecto_image_pipeline related to this one[1]?

[1]https://code.ros.org/trac/ros-pkg/ticket/5164

I'm using Indigo+ubuntu14.04(32bit)

I am trying to make some experiment with ORK（object recognition kitchen）.

But I've had a lot of trouble.Here are my steps follow the tutorial of ORK:

First : sudo apt-get install ros-indigo-ecto-*

Second : sudo apt-get install ros-indigo-object-recognition-*

Third : database configuration

Last : follow Quick Guide belows:

① rosrun object_recognition_core push.sh ② source /opt/indigo/setup/ros.sh ③ roscore ④roslaunch openni_launch openni.launch⑤rosrun object_recognition_capture orb_template -o my_textured_plane

Then the KeyError hanppend：
*exbot@ubuntu:~$ rosrun object_recognition_capture orb_template -o my_textured_plane
Traceback (most recent call last):
File "/opt/ros/indigo/lib/object_recognition_capture/orb_template", line 7, in
from ecto_opencv.calib import PoseDrawer, DepthValidDraw
KeyError: 'postregister'
*
In another way,I tried to build it from source,it's result like:
*exbot@ubuntu:~$ rosrun object_recognition_capture orb_template -o my_textured_plane
Listening for key: s on imshow:save
Registration? 1
Sync? 0
Segmentation fault (core dumped)
*
I don't know how to deal with it,Please give me a guide or some advice, thanks.

Hi,

I followed the install instructions for ros groovy, both from ros-repos and from source, but I get an error when running rosrun object_recognition_capture orb_template.py -o my_textured_plane:

Traceback (most recent call last): File "/home/stfn/devn/ros_dai_groovy/src/capture/apps/orb_template", line 6, in <module> from ecto_image_pipeline.io.source import add_camera_group, create_source ImportError: cannot import name add_camera_group

Any ideas? Sounds like an env issue to me. But I couldn't find

p.s. Is it normal, that I'm asked which file I mean?
`stfn@stfn-MacBook:~/devn/ros_dai_groovy/devel$ rosrun object_recognition_capture orb_template -o my_textured_plane
[rosrun] You have chosen a non-unique executable, please pick one of the following:

1. /home/stfn/devn/ros_dai_groovy/src/capture/doc/source/apps/orb_template
2. /home/stfn/devn/ros_dai_groovy/src/capture/apps/orb_template
3. /home/stfn/devn/ros_dai_groovy/src/capture/doc/source/apps/orb_template
4. /home/stfn/devn/ros_dai_groovy/src/capture/apps/orb_template`

Right now you can only capture from a plugged in sensor, it would be handy to be able to train from a topic (say, rosbag, running robot, whatever).

Also, the first image that comes out from my kinect/xtion is always super dark because of warming up of the device and that may be bad for the training to be done later on with the captured data.

When I (according to the documentation) first run
roslaunch openni_launch openni.launch and then
rosrun object_recognition_capture capture or orb_template, the scripts crash with the following error:

stfn@stfn-MacBook:~/devn/ros_dai_groovy_catkin$ rosrun object_recognition_capture orb_template -o tex_plane_poster
Listening for key: s on imshow:save
Registration? 1
Sync? 0
python: /usr/include/boost/smart_ptr/shared_ptr.hpp:412: boost::shared_ptr<T>::reference boost::shared_ptr<T>::operator*() const [with T = xn::NodeInfo, boost::shared_ptr<T>::reference = xn::NodeInfo&]: Assertion `px != 0' failed.
Aborted (core dumped)

I guess your source_node is trying to access the device even if openni_launch is already grabbing the images. Is the documentation outdated and you don't support grabbing images from ros-topics resp. openni_launch anymore?

Maybe using:
http://svn.pointclouds.org/pcl/trunk/gpu/kinfu_large_scale/tools/standalone_texture_mapping.cpp

Hi, I scanned my own template and run
rosrun object_recognition_capture capture -i my_textured_plane -o orc_scan_can_redbull.bag -m

The script crashes with the error

OpenCV Error: Assertion failed (!outImage.empty()) in drawKeypoints, file /tmp/buildd/ros-groovy-opencv2-2.4.3-6precise-20130125-0341/modules/features2d/src/draw.cpp, line 115
Traceback (most recent call last):
  File "/home/stfn/devn/ros_dai_groovy_catkin/src/wg_perception/capture/apps/capture", line 76, in <module>
    run_plasm(options, plasm, locals=dict(plasm=plasm, segmentation=segmentation))
  File "/home/stfn/devn/ros_dai_groovy_catkin/src/ecto_kitchen/ecto/python/ecto/opts.py", line 85, in run_plasm
    sched.execute(options.niter)
ecto.CellException:            exception_type  CellException
                cell_name  OrbPoseEstimator
                cell_type  ecto::py::BlackBox
            function_name  process
                     type  std::runtime_error
                     what             exception_type  CellException
                cell_name  DrawMatches
                cell_type  DrawMatches
            function_name  process
                     type  cv::Exception
                     what  /tmp/buildd/ros-groovy-opencv2-2.4.3-6precise-20130125-0341/modules/features2d/src/draw.cpp:115: error: (-215) !outImage.empty() in function drawKeypoints

terminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<boost::lock_error> >'
  what():  boost::lock_error
Aborted (core dumped)

I have two more issues about the scanning:
First, why does the script read a lena image:

Registered:on Supported: 1
Setting registration on
Setting sync off
Reading : my_textured_plane/T.yaml
Reading : my_textured_plane/R.yaml
Reading : my_textured_plane/descriptors.yaml
Reading : my_textured_plane/points3d.yaml
Reading : my_textured_plane/points.yaml
read image:lena.jpg
width:0 height:0 channels:1

And second: Without the m-param the capture script seems to work, the coordinate origin is updated when i move the template, but nothing happens. The mask image remains dark (what is it doing anyway?), and no metter how long i scan and turn, the script wont finish.

Any ideas?
Thanks!

When running
rosrun object_recognition_core detection -c src/wg_perception/tod/conf/detection.ros.ork
the script crashes with:

[ INFO] [1363632596.119472680]: Initialized ROS. node_name: /object_recognition_server
Traceback (most recent call last):
  File "/home/stfn/devn/ros_dai_groovy_catkin/src/wg_perception/object_recognition_core/apps/detection", line 21, in <module>
    plasm = create_plasm(ork_params)
  File "/home/stfn/devn/ros_dai_groovy_catkin/src/wg_perception/object_recognition_core/python/object_recognition_core/pipelines/plasm.py", line 66, in create_plasm
    cells[cell_name] = cells[cell_name](cell_name=cell_name, n_inputs=n_inputs, **ork_params[cell_name])
  File "/home/stfn/devn/ros_dai_groovy_catkin/src/wg_perception/object_recognition_core/python/object_recognition_core/io/voter.py", line 37, in __init__
    ecto.BlackBox.__init__(self, *args, **kwargs)
  File "/home/stfn/devn/ros_dai_groovy_catkin/src/ecto_kitchen/ecto/python/ecto/blackbox.py", line 248, in __init__
    self.__class__.__name__) + 'needs to be decorated with @classmethod or @staticmethod.')
ecto.blackbox.BlackBoxError: The "declare_cells" function of the BlackBox Aggregator needs to be decorated with @classmethod or @staticmethod.

With the other detection configuration file (detection.ork), I get a different error:

Registration? 1
Sync? 0
Warning: USB events thread - failed to set priority. This might cause loss of data...
1. device on bus: 1 @ 8 with serial number: A00364801973051A  Microsoft : Xbox NUI Camera
Time diff max: 33.5333
Registered:on Supported: 1
Setting registration on
Setting sync off
[FATAL] [1363632458.136088039]: You must call ros::init() before creating the first NodeHandle
[FATAL] [1363632458.136163239]: BREAKPOINT HIT
    file = /tmp/buildd/ros-groovy-roscpp-1.9.41-0precise-20130125-2016/src/libros/node_handle.cpp
    line=151

Trace/breakpoint trap (core dumped)

I'm following these tutorials: http://wg-perception.github.io/object_recognition_core/quickguide.html#quickguide
But when I run object_recognition_capture upload, I get this output:
Stored new object with id: 51679ad398cd1701a5182dae7a000549
Loading bag with : final.bag
running graph
Opening bag: final.bag
Inserting frame: 0
libpng warning: Image width is zero in IHDR
libpng warning: Image height is zero in IHDR
libpng error: Invalid IHDR data
OpenCV Error: Assertion failed (code) in imencode, file /build/buildd/opencv-2.4.8+dfsg1/modules/highgui/src/loadsave.cpp, line 430
Traceback (most recent call last):
File "/home/tomas/libs_and_graphics/ws/src/ork_capture/apps/upload", line 145, in
session = compute_for_bag_with_mask_pose(bag, obj, args)
File "/home/tomas/libs_and_graphics/ws/src/ork_capture/apps/upload", line 105, in compute_for_bag_with_mask_pose
run_plasm(args, plasm, locals=vars())
File "/home/tomas/libs_and_graphics/ws/src/ecto/python/ecto/opts.py", line 85, in run_plasm
sched.execute(options.niter)
ecto.CellException: exception_type CellException
[cell_name] = db_inserter

[cell_type] = object_recognition_core::prototypes::ObservationInserter

[function_name] = process_with_only_these_inputs

[type] = cv::Exception

[what] = /build/buildd/opencv-2.4.8+dfsg1/modules/highgui/src/loadsave.cpp:430: error: (-215) code in function imencode

While trying to follow the ORK tutorial, I hit a crash:

$ rosrun object_recognition_capture capture --seg_z_min 0.01 -o my_mug.bag
Traceback (most recent call last):
  File "/opt/ros/jade/lib/object_recognition_capture/capture", line 72, in <module>
    fps=options.fps
  File "/opt/ros/jade/lib/python2.7/dist-packages/object_recognition_capture/openni_capture.py", line 207, in create_capture_plasm
    plasm.connect(graph)
ecto.TypeMismatch:            exception_type  TypeMismatch
[to_key] = image

[to_cell] = Bag Writer if R|T

[from_key] = image

[from_cell] = ecto_ros::Mat2Image

[to_typename] = boost::shared_ptr<sensor_msgs::Image_<std::allocator<void> > const>

[from_typename] = boost::shared_ptr<sensor_msgs::Image_<std::allocator<void> > >

At first sight, the types look compatible to me (from non const to const should not be an issue, right?)...

Note that the same capture in preview mode, works well. ie,

$ rosrun object_recognition_capture capture --seg_z_min 0.01 -o my_mug.bag --preview

works as expected.

Found on Ubuntu 15.10, fresh ROS, manually compiled.

Hi community,
when I try to run rosrun object_recognition_capture capture --seg_z_min 0.01 -o OBJECT_NAME.bag --preview I get this error:

Registration? 1
Sync? 0
Traceback (most recent call last):
  File "/root/catkin_ws/src/capture/apps/capture", line 75, in <module>
    run_plasm(options, plasm, locals=dict(plasm=plasm, segmentation=segmentation))
  File "/opt/ros/indigo/lib/python2.7/dist-packages/ecto/opts.py", line 85, in run_plasm
    sched.execute(options.niter)
ecto.CellException:            exception_type  CellException
[cell_name] = Source

[cell_type] = ecto::py::BlackBox

[function_name] = process_with_only_these_inputs

[type] = std::runtime_error

[what] =            exception_type  CellException
[cell_name] = <class 'ecto_openni.OpenNICapture'>

[cell_type] = ecto_openni::OpenNICapture

[function_name] = configure

[type] = openni_wrapper::OpenNIException

[what] = openni_wrapper::OpenNIDevice::OpenNIDevice(xn::Context&, const xn::NodeInfo&, const xn::NodeInfo&, const xn::NodeInfo&, const xn::NodeInfo&) @ /tmp/binarydeb/ros-indigo-ecto-openni-0.4.0/openni_wrapper/src/openni_device.cpp @ 94 : creating depth generator failed. Reason: USB interface is not supported!

By default openni_launch is used, which works fine with the kinect camera. I want to use the Asus Xtion Pro Live, which needs openni2_launch. I checked the source code, but I couldn't find a flag or option to change this. I think this should be changed, not everyone uses the kinect.
Thank you for your help.

At some point in the quick guide you state:
'The currently stored models are on http://localhost:5984/model_viewer/_design/viewer/index.html'
but the URL only returns
{"error":"not_found","reason":"no_db_file"}

Hi,
I want to do capture -> upload -> mesh with my Intel realsense D415 camera.
Setup: Ubuntu 16.04, OpenCV 3.3.1, ROS kinetic
detection, orb_template, orb_track is working.

I adapted the capture code so that the capture process is working. Even when I get a segmentation dump at the very end, the rosbag is filled with all the needed topics. Moreover, I also saw the markers in the GUI were recognized properly when I turn the dot pattern.

So the next step for me is to upload this bag to the database, which I configured like mentioned in the documentation. It seems that the upload process loads something to the database(at least i can see a new object in the db), but I am not sure if this worked correctly, because I get the following output:

Inserting frame: 0 
libpng warning: Image width is zero in IHDR 
libpng warning: Image height is zero in IHDR 
libpng error: Invalid IHDR data 
OpenCV Error: Assertion failed (code) in imencode, file /tmp/binarydeb/ros-kinetic-opencv3-3.3.1/modules/imgcodecs/src/loadsave.cpp, line 899 
Traceback (most recent call last): 
  File "/opt/ros/kinetic/lib/object_recognition_capture/upload", line 145, in <module> 
    session = compute_for_bag_with_mask_pose(bag, obj, args) 
  File "/opt/ros/kinetic/lib/object_recognition_capture/upload", line 105, in compute_for_bag_with_mask_pose 
    run_plasm(args, plasm, locals=vars()) 
  File "/opt/ros/kinetic/lib/python2.7/dist-packages/ecto/opts.py", line 85, in run_plasm 
    sched.execute(options.niter) 
ecto.CellException:            exception_type  CellException 
[cell_name] = db_inserter 
[cell_type] = object_recognition_core::prototypes::ObservationInserter 
[function_name] = process_with_only_these_inputs 
[type] = cv::Exception 
[what] = /tmp/binarydeb/ros-kinetic-opencv3-3.3.1/modules/imgcodecs/src/loadsave.cpp:899: error: (-215) code in function imencode 

I saw that there was a solution for this, when using a kinect:
#23
So I adapt my encodings accordingly to the ROS topics of my camera. Those encodings can now be found in my rosbag. But still the same error.

Is there any Idea why this error is thrown? Could it be the ROS version itself 16.04? Or could it also be my openCV version 3.3.1?

Attached you can find my rosbag with just 3 views, to keep the size small. And the modified capture code.
02042020.bag.tar.gz
capture_realsense_code.tar.gz

Probably not the right place to ask, but is there anything like a public dataset with bags for training and bags for detection/test available?

I tried to capture objects myself and the reconstructed mesh looks good in MeshLab, but the detection results with the TOD detector are very bad although objects should be sufficiently textured.

I'm trying to capture objects for training and i dont get any masked image,although i have trained objects successfully several times before using the same set of commands as mentioned here: https://groups.google.com/forum/#!msg/object-recognition-kitchen/ijpfJdG9q8E/FPG4HtRBr4AJ

Here are a few snapshots of the "blank masked image":