search

wg-perception / linemod

An OR pipeline based on LINE-MOD from OpenCV
47 stars 56 forks source link

LINE-MOD crashes maybe due to libosmesa6? #18

Open theNerd247 opened 8 years ago

theNerd247 commented 8 years ago

Currently LINE-MOD is crashing with the following error: 

Traceback (most recent call last):
  File "/opt/ros/indigo/lib/object_recognition_core/detection", line 24, in <module>
    run_plasm(args, plasm)
  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] = pipeline1

[function_name] = process_with_only_these_inputs

[type] = std::bad_alloc

[what] = std::bad_alloc

[when] = While triggering param change callbacks

I ran the current master branch on another machine (Ubuntu 14.04.2) which hasn't been updated since August 2015 and linemod works fine. So then I grabbed all the packages that updated on my system and cross referenced it with the list of system dependencies that rosdep shows for object_recognition_linemod and libosmesa6-dev is the only package that is showing up. So I'm assuming a system update to libosmesa6-dev is causing the problem. However downgrading to libosmesa6-dev.10.1.0-4ubuntu5 doesn't seem to help anything.

nlyubova commented 8 years ago

Hello,

I'm using Ubuntu 14.04.3 and it works well for me. Did you try the last version of ecto, ecto_image_pipeline?

theNerd247 commented 8 years ago

Yes, here's what I ran:

$ sudo apt-get update
$ rosdep resolve ecto ecto_image_pipeline | grep -v "#" | xargs sudo apt-get install                                          
Reading package lists... Done
Building dependency tree       
Reading state information... Done
ros-indigo-ecto is already the newest version.
ros-indigo-ecto-image-pipeline is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 6 not upgraded.
nlyubova commented 8 years ago

ok, by the way, libosmesa6-dev is used by ork_renderer, did you get its last version as well?

theNerd247 commented 8 years ago

Yes they're at the newest version also. Are you using the ros-indigo-object-recognition-linemod package or are you building from the latest git source?

nlyubova commented 8 years ago

yeah I'm building from source all of these packages, but I will check the release and let you know By the way, which camera do you use?

theNerd247 commented 8 years ago

I'm using a kinect2 with the iai_kinect2 ros package.

theNerd247 commented 8 years ago

Any luck?

jontromanab commented 8 years ago

Have you solved the problem?? I am facing the exact same problem now

hris2003 commented 8 years ago

@theNerd247 Hi, can you try compiling ork_renderer (a dependency of LINEMOD) with GLUT instead of osmesa? Here is a discussion in ORK maillinglist about the same issue.

theNerd247 commented 8 years ago

Sadly I can't currently. The project that used linemod was with work and I won't be able to work on the same system till this summer.

YantianZha commented 8 years ago

Hi all, I met the same issue (https://drive.google.com/file/d/0BwZMlUaV88n6ODNockdLSVNxR1k/view?usp=sharing). I'm using Ubuntu 14.04.3 LTS, ROS indigo, and Kinect (with libfreenect driver). If you want more information please let me know. Thanks in advance for your help!

vrabaud commented 8 years ago

Can any of you please detail the different instructions you are launching ? For ORK or for the camera. Thx.

YantianZha commented 8 years ago

Because I'm using ROS, so the instructions are the ROS part on the ORK tutorial, and especially the linemod which got me the error: http://wg-perception.github.io/ork_tutorials/tutorial03/tutorial.html

The specific instruction I used which caused the problem is the detection part in linemod tutorial above, i.e., rosrun object_recognition_core detection -c rospack find object_recognition_linemod/conf/detection.ros.ork.

I met no problem before detection, with the given coke model and a coke.

Thanks!

theNerd247 commented 8 years ago

I am using exactly the same steps for running LINEMOD.

On January 1, 2016, at 6:03 PM, YantianZha notifications@github.com wrote:

Because I'm using ROS, so the instructions are the ROS part on the ORK tutorial, and especially the linemod which got me the error: http://wg-perception.github.io/ork_tutorials/tutorial03/tutorial.html

The specific instruction I used which caused the problem is the detection part in linemod tutorial above, i.e., rosrun object_recognition_core detection -c rospack find object_recognition_linemod/conf/detection.ros.ork.

I met no problem before detection, with the given coke model and a coke.

Thanks!

— Reply to this email directly or view it on GitHub.

YantianZha commented 8 years ago

And I'm sure my libfreenect driver ran ok. Because I got the right value from the four subscribed camera topics (you would see them after rosrun object_recognition_core detection -c rospack find object_recognition_linemod/conf/detection.ros.ork)

vrabaud commented 8 years ago

ell, I cannot reproduce with Jade on Vivid. Can you please trace down the error by using gdb ? You can do so with ecto like that: https://plasmodic.github.io/ecto/ecto/usage/external/debugging.html Don't use rosrun, just do:

gdb --args python ./object_recognition_core/apps/detection -c  ./object_recognition_linemod/conf/detection.ros.ork
YantianZha commented 8 years ago

The tutorial requires me to add the namespace of cell, but the cell name below is "pipeline1", which is actually from the conf file of detection (https://github.com/wg-perception/linemod/blob/master/conf/detection.ros.ork)? I'm new to these ecto/ork things. So can you give me more suggestions? Thanks!

Starting program: /usr/bin/python /home/yantian/ws/src/ork_core/apps/detection -c /home/yantian/ws/src/ork_linemod_new/conf/detection.ros.ork [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Traceback (most recent call last): File "/usr/share/gdb/auto-load/usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.19-gdb.py", line 63, in from libstdcxx.v6.printers import register_libstdcxx_printers ImportError: No module named 'libstdcxx' [New Thread 0x7fffd5196700 (LWP 16362)] [New Thread 0x7fffd3278700 (LWP 16368)] [New Thread 0x7fffd2a77700 (LWP 16369)] [New Thread 0x7fffd2276700 (LWP 16370)] [New Thread 0x7fffd1a75700 (LWP 16375)] [New Thread 0x7fffd1274700 (LWP 16376)] [New Thread 0x7fffd0a73700 (LWP 16377)] [New Thread 0x7fffc3fff700 (LWP 16378)] [New Thread 0x7fffc37fe700 (LWP 16379)] [ INFO] [1451694871.793249290]: Initialized ROS. node_name: /object_recognition_server [ INFO] [1451694874.324675346]: System already initialized. node_name: /object_recognition_server fffffffffffffffff Namespace(config_file='/home/yantian/ws/src/ork_linemod_new/conf/detection.ros.ork', dotfile='', graphviz=False, gui=False, ipython=False, logfile='', niter=0, stats=False, visualize=False) [New Thread 0x7fffabb97700 (LWP 16385)] [New Thread 0x7fffab396700 (LWP 16386)] [New Thread 0x7fffaab95700 (LWP 16387)] [New Thread 0x7fffaa394700 (LWP 16389)] [ INFO] [1451694874.332182778]: Subscribed to topic:/camera/depth_registered/camera_info [queue_size: 1][tcp_nodelay: 0] [Thread 0x7fffabb97700 (LWP 16385) exited] [ INFO] [1451694874.339498718]: Subscribed to topic:/camera/depth_registered/image_raw [queue_size: 1][tcp_nodelay: 0] [Thread 0x7fffaab95700 (LWP 16387) exited] [ INFO] [1451694874.345437435]: Subscribed to topic:/camera/rgb/image_rect_color [queue_size: 1][tcp_nodelay: 0] [Thread 0x7fffaa394700 (LWP 16389) exited] [ INFO] [1451694874.349263394]: Subscribed to topic:/camera/rgb/camera_info [queue_size: 1][tcp_nodelay: 0] [Thread 0x7fffab396700 (LWP 16386) exited] [New Thread 0x7fffaa394700 (LWP 16433)] [New Thread 0x7fffaab95700 (LWP 16434)] [New Thread 0x7fffab396700 (LWP 16435)] [New Thread 0x7fffabb97700 (LWP 16436)] [Thread 0x7fffabb97700 (LWP 16436) exited] Traceback (most recent call last): File "/home/yantian/ws/src/ork_core/apps/detection", line 27, in run_plasm(args, plasm) File "/home/yantian/ws/src/ecto/python/ecto/opts.py", line 86, in run_plasm sched.execute(options.niter) ecto.CellException: exception_type CellException [cell_name] = pipeline1

[function_name] = process_with_only_these_inputs

[type] = std::bad_alloc

[what] = std::bad_alloc

[when] = While triggering param change callbacks

[Thread 0x7fffd1a75700 (LWP 16375) exited] [Thread 0x7fffd3278700 (LWP 16368) exited] [Thread 0x7fffd2a77700 (LWP 16369) exited] [Thread 0x7fffc3fff700 (LWP 16378) exited] [Thread 0x7fffd1274700 (LWP 16376) exited] [Thread 0x7fffc37fe700 (LWP 16379) exited] [Thread 0x7fffd0a73700 (LWP 16377) exited] [Thread 0x7fffd5196700 (LWP 16362) exited] [Thread 0x7fffab396700 (LWP 16435) exited] [Thread 0x7fffaa394700 (LWP 16433) exited] [Thread 0x7ffff7fc5740 (LWP 16355) exited] [Thread 0x7fffd2276700 (LWP 16370) exited] Inferior 1 (process 16355) exited with code 01 break

hris2003 commented 8 years ago

@YantianZha Do you think you can provide us a rosbag of your camera's capture (point cloud and rgb and tf) so that we could debug LINEMOD with your setting? Thanks.

YantianZha commented 8 years ago

Thanks @hris2003 . Here is the link to my bag file: https://drive.google.com/file/d/0B6lGtF9IuZWCV2tzTHl6bXRFemM/view?usp=sharing The file is generated by running freenect.launch, roscore, and rqt_reconfigure (in ORK tutorial). Therefore you would see the registered pointcloud2 in RViz.

Then after training the coke model, using "rosrun object_recognition_core detection -c rospack find object_recognition_linemod/conf/detection.ros.ork", setting up "/camera_depth_optical_frame", you would see the error.

hris2003 commented 8 years ago

Hi,

So I run ork_linemod with your bag, and it works. No errors, no warning. Rviz shows the ork_objects detected in its window. Attached is a screenshot of that. One particular setting in my system is that ork_renderer is compiled with GLUT and not with osmesa.

For your problem, I suspect either something's wrong with the database or maybe ork_renderer is messing up. I recommend you to restart from zero, i.e.

  1. clear everything in couchdb,
  2. remove all compiled ork package,
  3. Configure ork_renderer so that it will be compiled with GLUT (instructions to do that can be found in another discussion of this maillinglist https://groups.google.com/d/msgid/object-recognition-kitchen/12546f19-4033-4ecf-a854-54d91bebf73d%40googlegroups.com ),
  4. recompile the source code of ork
  5. run the linemod training phrase
  6. run your camera's capture node
  7. run the linemod detection phrase

Good luck.

Ha Dang.

On Sat, Jan 2, 2016 at 6:22 PM, YantianZha notifications@github.com wrote:

Thanks @hris2003 https://github.com/hris2003 . Here is the link to my bag file: https://drive.google.com/file/d/0B6lGtF9IuZWCV2tzTHl6bXRFemM/view?usp=sharing The file is generated by running freenect.launch, roscore, and rqt_reconfigure (in ORK tutorial). Therefore you would see the registered pointcloud2 in RViz.

— Reply to this email directly or view it on GitHub https://github.com/wg-perception/linemod/issues/18#issuecomment-168409175 .

YantianZha commented 8 years ago

Some lines in https://groups.google.com/d/msgid/object-recognition-kitchen/12546f19-4033-4ecf-a854-54d91bebf73d%40googlegroups.com need to be changed:

Then go to src/CMakeLists.txt and block these few line in the header

if USE_GLUT

else

include "renderer3d_impl_osmesa.h"

endif

"Then go to src/CMakeLists.txt and block these few line in the header" --> " Then go to src/renderer3d.cpp and block these few line in the header". If I'm wrong please correct it.

hris2003 commented 8 years ago

For 6, I meant run the freenect.launch.

And yes, the concerned file is renderer3d.cpp (makhabhi@isu.edu did send a followup message to correct his instruction).

Ha Dang.

On Sat, Jan 2, 2016 at 10:10 PM, YantianZha notifications@github.com wrote:

"...5. run the linemod training phrase

  1. run your camera's capture node
  2. run the linemod detection phrase..."

Why I need to do " run your camera's capture node" ? Capture is used for creating the mesh ( http://wg-perception.github.io/ork_tutorials/tutorial01/tutorial.html#prepare-object-s-mesh)? And your step 5 training DO NEED a given mesh model as the input.

And the linemod usage tutorial mentions nothing about "run capture node" ( http://wg-perception.github.io/ork_tutorials/tutorial03/tutorial.html)

BTW: Some lines in https://groups.google.com/d/msgid/object-recognition-kitchen/12546f19-4033-4ecf-a854-54d91bebf73d%40googlegroups.com need to be changed:

Then go to src/CMakeLists.txt and block these few line in the header

if USE_GLUT

else

include "renderer3d_impl_osmesa.h"

endif

"Then go to src/CMakeLists.txt and block these few line in the header" --> " Then go to src/renderer3d.cpp and block these few line in the header". If I'm wrong please correct it.

— Reply to this email directly or view it on GitHub https://github.com/wg-perception/linemod/issues/18#issuecomment-168428334 .

YantianZha commented 8 years ago

I just did everything again, including reinstalling ecto. The error still emerged. Different from my previous trial using "mkdir ws && cd ws wstool init src https://raw.github.com/wg-perception/object_recognition_core/master/doc/source/ork.rosinstall cd src && wstool update -j8 cd .. && rosdep install --from-paths src -i -y catkin_make source devel/setup.bash" This time I install everything manually (otherwise, some ork packages would be installed in opt/ros/.../share folder which are incomplete.

Further, I met same error (ecto.cellException) in using "rosrun object_recognition_capture orb_template -o my_textured_plane", following http://wg-perception.github.io/object_recognition_core/quickguide.html#detect-objects. But considering that the linemod tutorial did not mention using "object_recognition_capture", I don't know whether this matters. Thanks!

jontromanab commented 8 years ago

I think it matters as it depends on by which library package you are creating your database in the training phase. (osmesa or glut). If you did not created your database with glut, then linemod would be generating the same error message you received. The linemod is not working currently with osmesa. If you have created your database by ork_renderer using osmesa, the tabletop will still work. U can give it a try.

Please refer to this page where I have tried to summarize how we can run the linemod. https://groups.google.com/forum/#!msg/object-recognition-kitchen/aS9YlKJxOnc/siFhK4uKDwAJ If you feel any problem, I will be happy to help (Also sorry for the mistake, I have already provided a follow-up message)

Thank you.

hris2003 commented 8 years ago

@YantianZha Do you think you can try ork on a different computer where there's not yet ork installed?

YantianZha commented 8 years ago

Yes it worked. Thank you so much all guys!

And @jontromanab , I'm sure I deleted the object firstly as what you suggested in your post. But I did not delete the overall data base (the mesh is still in it, delete object cmd can't delete the mesh)

I think the reason is, while I changed a list of files in renderer folder, and then ran catkin_make, my system still "make with the previous configuration". For example, you may note before that, if you just change a little part in a "certain place" in a catkin package, catkin_make would not actually "see" it, but just passed the make process.

Thanks again :)

YantianZha commented 8 years ago

Hi @jontromanab, Did you try try training more than one models (stl form) with glut? Did you meet "freeglut illegal glutInit() reinitialization attempt"? Thanks!

jontromanab commented 8 years ago

Hello @YantianZha ....Sorry for the late reply. Quite busy with the other stuffs. Yes. I think I have trained one coke can and one pringles chip box and it detected them separately. Though I am not quite sure, I will surely check it up tomorrow and let you know.

jontromanab commented 8 years ago

Hi @YantianZha Yes. I am having the same problem while training and trying to detect more than 1 object freeglut illegal glutInit() reinitialization attempt

Have you got any solution yet?

nlyubova commented 8 years ago

try OSMesa

catkin_make --cmake-args "-DUSE_GLUT=OFF"

frukostklubben commented 8 years ago

I have the exact problem described in this thread, I can't however run the training after compiling from source

training gives error: ----------------------------------------------------- Training 1 objects. computing object_id: b59867803c30d34e08e75141ea000329 Traceback (most recent call last): File "/home/roar/catkin_ws/src/object_recognition_core/apps/training", line 39, in plasm.execute() ecto.CellException: exception_type CellException [cell_name] = pipeline1

[cell_type] = ecto::py::BlackBox

[function_name] = process_with_only_these_inputs

[type] = std::runtime_error

[what] = exception_type CellException [cell_name] = <class 'object_recognition_linemod.ecto_cells.ecto_linemod.Trainer'>

[cell_type] = ecto_linemod::Trainer

[function_name] = process_with_only_these_inputs

[type] = std::runtime_error

[what] = value type is 172215184 not 2 -----------------------------------------------------

Sounds like there's a loop running wild. The bold value changes every time I try to run the program.

Any ideas on what might be the problem?

JimmyDaSilva commented 8 years ago

@theNerd247 , @jontromanab Did you guys ever succeeded in using ORK with the Kinectv2 and libfreenect2 ?

nosepas1 commented 8 years ago

@trykks Have you been able to resolve the issue? I'm having the same problem!

frukostklubben commented 8 years ago

I'm afraid not. I abandoned it completely due to lack of time in our project.

JimmyDaSilva commented 8 years ago

This is due to mesa issues that hasn't been solved yet. You can use Glut instead, just change the CMakelist.txt like this : https://github.com/JimmyDaSilva/ork_renderer/commit/4bdd53e3c418e7d02be0212ece04598619b4323a

nosepas1 commented 8 years ago

@JimmyDaSilva I've done what you said and it doesn't work! do you have any other suggestion?

JimmyDaSilva commented 8 years ago

Well what is the error exactly?

nosepas1 commented 8 years ago

the error is this one:

Training 2 objects. computing object_id: 7a5b1640af7fd6e5b6fc990f45000b2c Traceback (most recent call last): File "/home/marti/catkin_ws/src/object_recognition_core/apps/training", line 39, in plasm.execute() ecto.CellException: exception_type CellException [cell_name] = pipeline1

[cell_type] = ecto::py::BlackBox

[function_name] = process_with_only_these_inputs

[type] = std::runtime_error

[what] = exception_type CellException [cell_name] = <class 'object_recognition_linemod.ecto_cells.ecto_linemod.Trainer'>

[cell_type] = ecto_linemod::Trainer

[function_name] = process_with_only_these_inputs

[type] = std::runtime_error

[what] = value type is 50032384 not 2

that's when I'm trying to train the objects.

2016-05-04 10:50 GMT+01:00 Jimmy DA SILVA notifications@github.com:

Well what is the error exactly?

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/wg-perception/linemod/issues/18#issuecomment-216815281

JimmyDaSilva commented 8 years ago

I am no expert here since the error doesn't help a lot but I can try to help.

Can you please:

And tell me which training fails. The problem could come from different sources. Maybe the meshes are wrong, maybe it's a glut issue, ... can't really tell until you try this.

nosepas1 commented 8 years ago

Now it's working. I would be able to train the objects. I install it from the repositories and the training works. Then, what I try first is do the detection using the source and I get a diferent error that it's is no from ork, it's from the opencv. The error is this one:

[ INFO] [1462362900.067559485]: Initialized ROS. node_name: /object_recognition_server [ INFO] [1462362901.970808495]: System already initialized. node_name: /object_recognition_server [ INFO] [1462362901.980498554]: Subscribed to topic:/camera/depth_registered/camera_info [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.989085679]: Subscribed to topic:/camera/rgb/image_rect_color [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.993443985]: Subscribed to topic:/camera/rgb/camera_info [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.997705420]: Subscribed to topic:/camera/depth_registered/image_raw [queue_size: 1][tcp_nodelay: 0] OpenCV Error: Bad argument (Invalid pointer to file storage) in cvGetRootFileNode, file /home/marti/ork/src/opencv-2.4.12/modules/core/src/persistence.cpp, line 803

do you know what's happening here?

thanks in advance

2016-05-04 12:43 GMT+01:00 Jimmy DA SILVA notifications@github.com:

I am no expert here since the error doesn't help a lot but I can try to help.

Can you please:

  • remove all the objects from the database
  • add the coke can from the tutorials to the database
  • try training
  • remove all objects again
  • add one of your objects
  • try training
  • add the coke can again to be the second object available
  • try training

And tell me which training fails. The problem could come from different sources. Maybe the meshes are wrong, maybe it's a glut issue, ... can't really tell until you try this.

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/wg-perception/linemod/issues/18#issuecomment-216836224

nosepas1 commented 8 years ago

by the way, the training from the source it's is not working, I'm having the same error. I'll try what you've said.

2016-05-04 12:58 GMT+01:00 Marti de castro cros < marti.de.castro.cros@gmail.com>:

Now it's working. I would be able to train the objects. I install it from the repositories and the training works. Then, what I try first is do the detection using the source and I get a diferent error that it's is no from ork, it's from the opencv. The error is this one:

[ INFO] [1462362900.067559485]: Initialized ROS. node_name: /object_recognition_server [ INFO] [1462362901.970808495]: System already initialized. node_name: /object_recognition_server [ INFO] [1462362901.980498554]: Subscribed to topic:/camera/depth_registered/camera_info [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.989085679]: Subscribed to topic:/camera/rgb/image_rect_color [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.993443985]: Subscribed to topic:/camera/rgb/camera_info [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.997705420]: Subscribed to topic:/camera/depth_registered/image_raw [queue_size: 1][tcp_nodelay: 0] OpenCV Error: Bad argument (Invalid pointer to file storage) in cvGetRootFileNode, file /home/marti/ork/src/opencv-2.4.12/modules/core/src/persistence.cpp, line 803

do you know what's happening here?

thanks in advance

2016-05-04 12:43 GMT+01:00 Jimmy DA SILVA notifications@github.com:

I am no expert here since the error doesn't help a lot but I can try to help.

Can you please:

  • remove all the objects from the database
  • add the coke can from the tutorials to the database
  • try training
  • remove all objects again
  • add one of your objects
  • try training
  • add the coke can again to be the second object available
  • try training

And tell me which training fails. The problem could come from different sources. Maybe the meshes are wrong, maybe it's a glut issue, ... can't really tell until you try this.

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/wg-perception/linemod/issues/18#issuecomment-216836224

nosepas1 commented 8 years ago

I'm having the same error everytime I try to train.

Training 1 objects. computing object_id: ec724c057ef20fc889510b60de003ca8 Traceback (most recent call last): File "/home/marti/ork/src/object_recognition_core/apps/training", line 39, in plasm.execute() ecto.CellException: exception_type CellException [cell_name] = pipeline1

[cell_type] = ecto::py::BlackBox

[function_name] = process_with_only_these_inputs

[type] = std::runtime_error

[what] = exception_type CellException [cell_name] = <class 'object_recognition_linemod.ecto_cells.ecto_linemod.Trainer'>

[cell_type] = ecto_linemod::Trainer

[function_name] = process_with_only_these_inputs

[type] = std::bad_alloc

[what] = std::bad_alloc

any other ideas?

thanks

2016-05-04 13:00 GMT+01:00 Marti de castro cros < marti.de.castro.cros@gmail.com>:

by the way, the training from the source it's is not working, I'm having the same error. I'll try what you've said.

2016-05-04 12:58 GMT+01:00 Marti de castro cros < marti.de.castro.cros@gmail.com>:

Now it's working. I would be able to train the objects. I install it from the repositories and the training works. Then, what I try first is do the detection using the source and I get a diferent error that it's is no from ork, it's from the opencv. The error is this one:

[ INFO] [1462362900.067559485]: Initialized ROS. node_name: /object_recognition_server [ INFO] [1462362901.970808495]: System already initialized. node_name: /object_recognition_server [ INFO] [1462362901.980498554]: Subscribed to topic:/camera/depth_registered/camera_info [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.989085679]: Subscribed to topic:/camera/rgb/image_rect_color [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.993443985]: Subscribed to topic:/camera/rgb/camera_info [queue_size: 1][tcp_nodelay: 0] [ INFO] [1462362901.997705420]: Subscribed to topic:/camera/depth_registered/image_raw [queue_size: 1][tcp_nodelay: 0] OpenCV Error: Bad argument (Invalid pointer to file storage) in cvGetRootFileNode, file /home/marti/ork/src/opencv-2.4.12/modules/core/src/persistence.cpp, line 803

do you know what's happening here?

thanks in advance

2016-05-04 12:43 GMT+01:00 Jimmy DA SILVA notifications@github.com:

I am no expert here since the error doesn't help a lot but I can try to help.

Can you please:

  • remove all the objects from the database
  • add the coke can from the tutorials to the database
  • try training
  • remove all objects again
  • add one of your objects
  • try training
  • add the coke can again to be the second object available
  • try training

And tell me which training fails. The problem could come from different sources. Maybe the meshes are wrong, maybe it's a glut issue, ... can't really tell until you try this.

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/wg-perception/linemod/issues/18#issuecomment-216836224

JimmyDaSilva commented 8 years ago

So are you using ORK from the sources or from the release ? If using sources, using GLUT or Mesa ?

nosepas1 commented 8 years ago

I've try in both ways. From release it works properly. Then, I have try to run it from the source and I get all the these errors. And I'm using GLUT.

2016-05-04 15:28 GMT+01:00 Jimmy DA SILVA notifications@github.com:

So are you using ORK from the sources or from the release ? If using sources, using GLUT or Mesa ?

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/wg-perception/linemod/issues/18#issuecomment-216882379

JimmyDaSilva commented 8 years ago

Alright keep using the sources then. Are you sure you changed the CMakelist.txt of ork_renderer according to this : https://github.com/JimmyDaSilva/ork_renderer/commit/4bdd53e3c418e7d02be0212ece04598619b4323a

nosepas1 commented 8 years ago

yes, I'm sure about that. I have been looking for the error and it seems that it's something related with the inputs, maybe somthing related with the mesh but I've already try to delete all the objects and add them again, to reinstall the database and nothing works.

2016-05-04 15:52 GMT+01:00 Jimmy DA SILVA notifications@github.com:

Alright keep using the sources then. Are you sure you changed the CMakelist.txt of ork_renderer according to this : JimmyDaSilva/ork_renderer@4bdd53e https://github.com/JimmyDaSilva/ork_renderer/commit/4bdd53e3c418e7d02be0212ece04598619b4323a

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/wg-perception/linemod/issues/18#issuecomment-216889697

JimmyDaSilva commented 8 years ago

To find all the objects still existing: rosrun object_recognition_core object_search.py Then remove them with: rosrun object_recognition_core object_delete.py OBJECT_ID

That way we will make sure everything has been deleted (mesh, trained data, etc...) Then add the coke.stl from the tutorials and trying again

ccqedq commented 5 years ago

I met the same problem,line 24, in run_plasm(args, plasm) line 85, in run_plasm, sched.execute(options.niter) [what]=/tmp/binarydeb/ros-kinetic-opencv3-3.3.1/opencv_contrib/rgbd/src/linemod.cpp:1098:error:(-215) response_map.row%T==0 in function linearize, It seems that someone have solved it ,but It's still a problem for me

zhywhu commented 5 years ago

@theNerd247 @nlyubova @hris2003 @YantianZha hello, I met the same problem too.It was ok when i use the coke.stl, but when i use my own mesh file, erro happened: ` File "/home/zhy/ork_ws/src/object_recognition_core/apps/detection", line 24, in run_plasm(args, plasm) File "/opt/ros/kinetic/lib/python2.7/dist-packages/ecto/opts.py", line 85, in run_plasmsched.execute(options.niter) ecto.CellException: exception_type CellException

[cell_name] = pipeline1

[function_name] = process_with_only_these_inputs

[type] = std::bad_alloc

[what] = std::bad_alloc

[when] = While triggering param change callbacks ` (I created my own object with blender)