wg-perception / linemod Goto Github PK

linemod_train.cpp, Line 198:

float distance = fabs(renderer_iterator.D_obj() - float(depth.at<ushort>(depth.rows/2.0f, depth.cols/2.0f)/1000.0f));

use of fabs() is incorrect, since for some objects (eg. a bowl) the distance from (0,0,0) to the object surface along the camera line of sight may be negative. Just removing fabs() seems to work.

Hi
I am having a problem of pose deviation error. I there any possible way to correct the error. For example the pose for Coke can is above the coke can.

In linemode detector.py seems that requested Detector method is not implemented (anymore?) in linemod_detect.cpp.

rosrun object_recognition_core training -c rospack find object_recognition_linemod/conf/training.ork
Training 2 objects.
computing object_id: 26fd4a3e0309ca93c5d8ba054b001268
Info, T0: Load /tmp/filejUzsrU.obj
Info, T0: Found a matching importer for this file format
Info, T0: Import root directory is '/tmp/'
Error, T0: OBJ: Unable to locate material file paper_cup.mtl
Error, T0: OBJ: failed to locate material None.001, skipping
Info, T0: Entering post processing pipeline
Info, T0: Points: 0, Lines: 0, Triangles: 1, Polygons: 0 (Meshes, X = removed)
Error, T0: Failed to compute tangents; need UV data in channel0
Info, T0: JoinVerticesProcess finished | Verts in: 71400 out: 48417 | ~32.2%
Info, T0: Cache relevant are 1 meshes (23800 faces). Average output ACMR is 2.034622
Info, T0: Leaving post processing pipeline
Deleting the previous model 26fd4a3e0309ca93c5d8ba054b0027bb of object 26fd4a3e0309ca93c5d8ba054b001268
computing object_id: bd469776b362b00f0f60e3e3bd000425
freeglut illegal glutInit() reinitialization attempt

The first possible reason is: my model is wrong. For your convenience here is the link to my model: https://drive.google.com/folderview?id=0BwZMlUaV88n6OWNqbUtGUE84Y1U&usp=sharing

That said, I created the model with blender following the tutorial (http://wg-perception.github.io/ork_tutorials/tutorial01/tutorial.html#creating-an-object-in-the-db). There're two points: origin, and size. I think my model in right by following the coke.obj (NOT coke.stl) model as an sample.

Another possible reason is: #3
That link refers to wg-perception/ork_renderer@bf49f05
But I can't find the fie "src/renderer_glut.cpp". The closest file I can find is: "ork_renderer/src/renderer3d_impl_glut.cpp". But I suspect I don't need to change that file any more?

Thanks very much!

When ever i run the linemod detection I get this

Reading the mesh file original.obj
Info, T0: Load ���W��.obj
Info, T0: Found a matching importer for this file format
Info, T0: Import root directory is './'
Error, T0: OBJ: Unable to locate material file ./coke.obj.mtl
Error, T0: OBJ: failed to locate material material_0, skipping
Error, T0: OBJ: failed to locate material material_1, skipping
Error, T0: OBJ: failed to locate material material_2, skipping
Info, T0: Entering post processing pipeline
Info, T0: Points: 0, Lines: 0, Triangles: 3, Polygons: 0 (Meshes, X = removed)
Info, T0: CalcTangentsProcess finished. Tangents have been calculated
Info, T0: JoinVerticesProcess finished | Verts in: 1536 out: 478 | ~68.9%
Info, T0: Cache relevant are 3 meshes (512 faces). Average output ACMR is 1.005859
Info, T0: Leaving post processing pipeline
Loaded 539d95f3026597a4f48fe9bfd50009ff with the number of samples 4579

[ INFO] [1481820107.500078438]: publishing to topic:/recognized_object_array

but it never detects a coke can on my desk. I am using Kinect v2 and iai kinect bridge. Can anyone please help me.

Hi All
I compiled the ork_render with Glut and was trying the linemode training
curently the assimp render winodw is coming empty (transparent)during trainig and the detection afterwards seems very poor
any suggetions what might be the problem , and is the assimp render windows supposed to show the model during thr training
I am using kinect 1 and freenect
Cheers
Khaled

it seams that objects in /recognized_object_array have the same frame_id as an input RGB image even if we do not use RGB image. SO we can use only registered images (aligned depth and rgb), otherwise the transformation between RGB and depth topics is not taken into account

~/ork/src/ork_linemod/conf$ rosrun object_recognition_core training -c training.ork
Training 6 objects.
computing object_id: 8b8ec44926f9fec2c30d2c54b906ac40
Info, T0: Load
Error, T0: Unable to open file "".
Segmentation fault (core dumped)

I get this error when I try to train.

Then obviously if I try to detect objects I get:
No models found for method "LINEMOD"

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.

Hi,

I use the ORK LineMOD pipeline with a Intel Realsense d415 sensor at 640 x 480 pixels.

When detecting the tray on the images below, the detected coordinates are always somewhat off by a semi-consistent amount.


The window with the colored dots works normally. If I am not wrong this is the detection before performing ICP. The RVIZ window shows the resulting recognized_object_array. If I am not wrong this is the detection after performing ICP.
I checked in blender if the object model was centered properly so this is not the problem.

Does anyone have an idea how this could be caused?

To get ORK to work with the Realsense d415 I had to change some parameters like focal length and image topics. I use Realsense_camera instead of OpenNi to get the pointclouds (not outputting to pointcloud2 but to separate depth/image topics).

If anyone could help me out I that would be great.
Simon van Eeden

HI All

I am trying the object detection library for the first
I went through the tutorial and managed to get the tabletop example to work with freenect and kinect 1, currently when I run the Linemode detection after the training , my pc slows down till it becomes unresponsive without any errors in the terminal
I was wondering if anyone have faced this problem before

I am using the Ubuntu 14.04.5 with ROS indigo
Cheers
Khaled

I hope there is someone out there, how is able to fill in the missing pieces in the documentation: http://wg-perception.github.io/linemod/index.html#line-mod

I would love to give this detector/recogniser a try, but since I am newbie in the object detection world, I'm kind of lost without this tutorial. :-(

Thanks!

When the packet is installed : a bunch of ecto package are installe (ecto-opencv, ecto-ros, ecto-image-pipeline).
But when trying to launch sample/sample.py, the module ecto-openni is missing.
Should the dependency added here in package.xml? in ecto package ?

I followed the linemod tutorial step by step. however there always no publishing topics. The output as below:

ast@ast:~$ rosrun object_recognition_core detection -c rospack find object_recognition_linemod/conf/detection.ros.ork
[ INFO] [1462421591.079412378]: Initialized ROS. node_name: /object_recognition_server
[ INFO] [1462421591.589505137]: System already initialized. node_name: /object_recognition_server
[ INFO] [1462421591.612506094]: Subscribed to topic:/camera/rgb/image_rect_color [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1462421591.632331203]: Subscribed to topic:/camera/depth_registered/camera_info [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1462421591.641620751]: Subscribed to topic:/camera/depth_registered/image_raw [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1462421591.654347153]: Subscribed to topic:/camera/rgb/camera_info [queue_size: 1][tcp_nodelay: 0]

while the training process outputs as below, is there some error?

ast@ast:~$ rosrun object_recognition_core training -c rospack find object_recognition_linemod/conf/training.ork
Training 1 objects.
computing object_id: 8084c0c1974909d0519e05e401006d09
Info, T0: Load /tmp/fileMj3UPW.obj
Info, T0: Found a matching importer for this file format
Info, T0: Import root directory is '/tmp/'
Error, T0: OBJ: No object detected to attach a new mesh instance.
Skipping one or more lines with the same contents
Info, T0: Entering post processing pipeline
Warn, T0: Found 32 degenerated primitives
Info, T0: TriangulateProcess finished. All polygons have been triangulated.
Info, T0: Points: 0, Lines: 1, Triangles: 3, Polygons: 0 (Meshes, X = removed)
Info, T0: Normal vectors are undefined for line and point meshes
Info, T0: GenVertexNormalsProcess finished. Vertex normals have been calculated
Info, T0: Tangents are undefined for line and point meshes
Info, T0: CalcTangentsProcess finished. Tangents have been calculated
Info, T0: JoinVerticesProcess finished | Verts in: 1472 out: 461 | ~68.7%
Error, T0: This algorithm works on triangle meshes only
Info, T0: Cache relevant are 3 meshes (512 faces). Average output ACMR is 0.890625
Info, T0: Leaving post processing pipeline
texture loaded.
texture loaded.
texture loaded.
Texture reused.

Anybody can help?

Hi community,
after Linemod detected an object, I would like to read the exact coordinates of the detected object.
Something like:
P1(220/230)
P2(221/230)
...
I have found code that returns the bounding box of the detected object, but there are to many points that doesn't belong to the object. I tried to calculate the points with the R-matrix and T-matrix but I don't get it right. Can someone please explain me how I can calculate/get every point coordinate? Sorry if this is a stupid question, but I think those coordinates would be nice for everybody who tries to modify/improve LineMod. Thank you very much, I really appreciate it.

My output of every variable:

depth_real_model_raw: 67 40
depth_real_ref_raw: 640 480
Match: 87 257
Rect_ref: 87 257
Rect_model: 87 257 67 40
T: [-0.280007, 0.0317065, 0.189323]
 R: [0.16856061, -0.23157644, -0.95810217;
  -0.89435035, -0.44457686, -0.049888521;
  -0.41439685, 0.86528826, -0.28204873]
T: [-0.280007, 0.0317065, 0.189323]
 R: [0.16856061, -0.23157644, -0.95810217;
  -0.89435035, -0.44457686, -0.049888521;
  -0.41439685, 0.86528826, -0.28204873]
pts_real_ref_temp: 0 [-0.227042, 0.00267829, 0.546]
pts_real_model_temp: 0 [0.100638, -0.00946573, -0.257265]
pts_real_ref_temp: 1 [-0.225593, 0.00267338, 0.545]
pts_real_model_temp: 1 [0.100638, -0.00946573, -0.257265]
pts_real_ref_temp: 2 [-0.22456, 0.00267338, 0.545]
pts_real_model_temp: 2 [0.100638, -0.00946573, -0.257265]
pts_model.size: 2551
pts_model.1: [0.106741, -0.00810874, -0.264011]
pts_ref.1: [-0.225593, 0.00267338, 0.545]
it_o2-t: [-0.196518, 0.0211954, 0.554]
T: [-0.196662, 0.0221105, 0.553999]
 R: [0.2549791, -0.31554195, -0.91401243;
  -0.72558904, -0.68724585, 0.034840584;
  -0.6391449, 0.65431392, -0.40418717]
43af9b1708c74566067abf1a4801e645 94.0476 icp 0.260918,  ratio 0 or 0
matches  3 / 1 / 3

Linemod in opencv is using a linearMemoryPyramid with levels {5,8} by default:
https://github.com/Itseez/opencv_contrib/blob/master/modules/rgbd/src/linemod.cpp#L1834:L1840

5 and 8 work fine for 640x480 images, but fails with every image sizes provided by the KinectV2 (512x424, 960x540, 1920x1080).
The problem has been discussed there:
opencv/opencv#4593

So I am trying to not use getDefaultLINEMOD but create my own detector using these lines in linemod_detect.cpp:

static const int T_LVLS[] = {3, 4};
std::vector< cv::Ptr<cv::linemod::Modality> > modalities;
modalities.push_back(new cv::linemod::ColorGradient());
modalities.push_back(new cv::linemod::DepthNormal());
detector_ = new cv::linemod::Detector(modalities, std::vector<int>(T_LVLS, T_LVLS +2));

instead of
detector_ = getDefaultLINEMOD()

For some image sizes the assertion then passes and the code runs. Unfortunately no objects are found...

@vrabaud I guess you know a bit more about the image processing behind. Do think this is something that actually work? Which values for T would you pick? How many pyramid levels do you think are necessary ?

Thanks for the help !
Jimmy

hello everyone
Im now using linemod to detect the coke can with the ORK tutorials after the linemod training when I run the detection it cant detect anything in rviz.

grxuser@grxuser-XPS-8900:~/catkin_ws/src/ork_tutorials/data$ rosrun object_recognition_core detection -c rospack find object_recognition_linemod/conf/detection.ros.ork
[ INFO] [1511763503.230235142]: Initialized ROS. node_name: /object_recognition_server
[ INFO] [1511763503.511648858]: System already initialized. node_name: /object_recognition_server
[ INFO] [1511763503.516849406]: Subscribed to topic:/camera/depth_registered/sw_registered/camera_info [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1511763503.518889503]: Subscribed to topic:/camera/rgb/camera_info [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1511763503.520862192]: Subscribed to topic:/camera/rgb/image_rect_color [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1511763503.523170545]: Subscribed to topic:/camera/depth_registered/sw_registered/image_rect_raw [queue_size: 1][tcp_nodelay: 0]
Reading the mesh file original.stl
Info, T0: Load �<[+��.stl
Info, T0: Found a matching importer for this file format
Info, T0: Import root directory is './'
Info, T0: Entering post processing pipeline
Info, T0: Points: 0, Lines: 0, Triangles: 1, Polygons: 0 (Meshes, X = removed)
Error, T0: FindInvalidDataProcess fails on mesh normals: Found zero-length vector
Info, T0: FindInvalidDataProcess finished. Found issues ...
Info, T0: GenVertexNormalsProcess finished. Vertex normals have been calculated
Error, T0: Failed to compute tangents; need UV data in channel0
Info, T0: JoinVerticesProcess finished | Verts in: 1536 out: 258 | ~83.2%
Info, T0: Cache relevant are 1 meshes (512 faces). Average output ACMR is 0.669922
Info, T0: Leaving post processing pipeline
Loaded 82507099cc7cc8eedf2a98328f000009 with the number of samples 5100

[ INFO] [1511763504.639047482]: publishing to topic:/recognized_object_array

and when I check the published topic /recognized_object_arraygrxuser@grxuser-XPS-8900:~$ rostopic echo /recognized_object_array
header:
seq: 3450
stamp:
secs: 1511763845
nsecs: 235033047
frame_id: camera_rgb_optical_frame
objects: []
cooccurrence: []

header:
seq: 3451
stamp:
secs: 1511763845
nsecs: 335025021
frame_id: camera_rgb_optical_frame
objects: []
cooccurrence: []

I am using the latest version of ORK, Ros indigo, Ubuntu 14.04

Thank you very much

Hi, I am having trouble detecting the tutorial's coke can.

The /real_icpin_ref seems to have the coke can + a single point (See the green dot in the images). I don't really understand where this dot comes from.

I've been trying to run the linemod detection tutorial found here: http://wg-perception.github.io/ork_tutorials/tutorial03/tutorial.html

Training phase works fine, but the detection phase throws the following error:

-----------------------------------------------------------------------------------------------------------

[ INFO] [1434356493.822010980]: Initialized ROS. node_name: /object_recognition_server
[ INFO] [1434356494.100826452]: System already initialized. node_name: /object_recognition_server
[ INFO] [1434356494.108034166]: Subscribed to topic:/camera/depth_registered/camera_info with queue size of 1
[ INFO] [1434356494.111847947]: Subscribed to topic:/camera/rgb/camera_info with queue size of 1
[ INFO] [1434356494.118495788]: Subscribed to topic:/camera/rgb/image_rect_color with queue size of 1
[ INFO] [1434356494.123829464]: Subscribed to topic:/camera/depth_registered/image_raw with queue size of 1
Traceback (most recent call last):
File "/home/astrostudent/ork/src/object_recognition_core/apps/detection", line 24, in
run_plasm(args, plasm)
File "/home/astrostudent/ork/src/ecto/python/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

-----------------------------------------------------------------------------------------------------------

Any ideas on how to fix this?

Hi all,

There's a nodehandle in linemod_detect.cpp. https://github.com/wg-perception/linemod/blob/master/src/linemod_detect.cpp

This nodehandle (node_) would be used to construct linemod_pointcloud module. But I'm wondering where is the node that matches this nodehandle?

And a strange thing is, when I used the same "node_" to construct my own publisher, and when I rostopic echo the topic of my own publisher based on "node_", I got error "You must call ros::init() before creating the first NodeHandle"? If the nodehandle "node_" can work with linemod_pointcloud publisher, why my own publisher cannot work?

Thanks very much in advance!

Hi,
I use Linemod to detect 2 objects: a cup and a coke. I trained their CAD models using linemod too. However when I run detection command and visualize using Rviz, there are a lot of wrong results popping up (see photo below). When I use tabletop, everything works fine (only 2 object detected in the scene). Does anyone know what is wrong?

When I try to detect objects I get the following output and nothing happens after this:

rosrun object_recognition_core detection -c  `rospack find object_recognition_linemod`/conf/detection.ros.ork
[ INFO] [1463665199.331471619]: Initialized ROS. node_name: /object_recognition_server
[ INFO] [1463665199.714861103]: System already initialized. node_name: /object_recognition_server
[ INFO] [1463665199.724349959]: Subscribed to topic:/camera/depth_registered/camera_info [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1463665199.727949582]: Subscribed to topic:/camera/rgb/camera_info [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1463665199.731266661]: Subscribed to topic:/camera/depth_registered/image_raw [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1463665199.734579789]: Subscribed to topic:/camera/rgb/image_rect_color [queue_size: 1][tcp_nodelay: 0]

It never publishes something. After like 2 min of waiting the command get the following error message:

[ INFO] [1463665094.353870696]: Initialized ROS. node_name: /object_recognition_server
[ INFO] [1463665094.716635842]: System already initialized. node_name: /object_recognition_server
[ INFO] [1463665094.729798651]: Subscribed to topic:/camera/depth_registered/camera_info [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1463665094.738068257]: Subscribed to topic:/camera/depth_registered/image_raw [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1463665094.747247293]: Subscribed to topic:/camera/rgb/camera_info [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1463665094.754698440]: Subscribed to topic:/camera/rgb/image_rect_color [queue_size: 1][tcp_nodelay: 0]
Traceback (most recent call last):
  File "/home/marcel/catkin_ws/src/object_recognition_core/apps/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

This is my output from training:

rosrun object_recognition_core training -c `rospack find object_recognition_linemod`/conf/training.ork 
Training 1 objects.
computing object_id: 8e7cd947fb153e19c0437207ec001034
Info,  T0: Load /tmp/fileTzFANj.obj
Info,  T0: Found a matching importer for this file format
Info,  T0: Import root directory is '/tmp/'
Error, T0: OBJ: No object detected to attach a new mesh instance.
Skipping one or more lines with the same contents
Info,  T0: Entering post processing pipeline
Info,  T0: RemoveRedundantMatsProcess finished. Removed 0 redundant and 1 unused materials.
Info,  T0: Points: 0, Lines: 0, Triangles: 3, Polygons: 0 (Meshes, X = removed)
Info,  T0: CalcTangentsProcess finished. Tangents have been calculated
Info,  T0: JoinVerticesProcess finished | Verts in: 1536 out: 478 | ~68.9%
Info,  T0: Cache relevant are 3 meshes (512 faces). Average output ACMR is 1.005859
Info,  T0: Leaving post processing pipeline
texture loaded.
texture loaded.
texture loaded.
Deleting the previous model 8e7cd947fb153e19c0437207ec001ea3 of object 8e7cd947fb153e19c0437207ec001034

I also saw #30, but I am using the tutorial coke.stl

I am using the latest version of ORK, Ros indigo, Ubuntu 14.04 and OpenCV 2.4.11

Thank you very much

However, if I name a specific object_id, I do not get that warning:

object_ids: ['a27b36f95952f2772662a15fc0002acd']

I'm on the most recent commit, 5d06f07, Ubuntu 16.04, ROS Kinetic. Thank you, this looks like a great package.

Hi,

we are trying to use linemod to detect a white mug.
After training with the mesh model from https://3dwarehouse.sketchup.com/model/8a6712961d81f6843a8e51534d260c09/popular-mug-IKEA-F%C3%84RGRIK (we have modified the model since there are two mugs on it), linemod can only find the mug randomly and if it does, only from a camera distance larger than 1m. We tried to rescale it to a similar size of the coke.stl from the tutorial, but it worked even worse. Do you have any idea why it doesn't work?

The texture of the coke.obj model used in the Tutorials does not appear during training when I turn on LINEMOD_VIZ_IMG... Relevant code for visualization (linemod_train.cpp):

#if LINEMOD_VIZ_IMG
// Display the rendered image
if (*visualize_)
{
  cv::namedWindow("Rendering");
  if (!image.empty()) {
  cv::imshow("Rendering", image);
  cv::waitKey(1);
  }
}
#endif

I would like to use texture (including color) to improve detection of some objects...

Alberto.

Hi everyone,

when I try to run the detect code following official guide, there are some problems I meet:

yuhan@ubuntu:~/ork_ws$ rosrun object_recognition_core detection -c  `rospack find object_recognition_linemod`/conf/detection.ros.ork
[ INFO] [1646498078.276179011]: Initialized ROS. node_name: /object_recognition_server
[ INFO] [1646498078.621038243]: System already initialized. node_name: /object_recognition_server
[ INFO] [1646498078.628744160]: Subscribed to topic:/camera/color/camera_info [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1646498078.633553914]: Subscribed to topic:/camera/depth/image_rect_raw [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1646498078.637222697]: Subscribed to topic:/camera/color/image_raw [queue_size: 1][tcp_nodelay: 0]
[ INFO] [1646498078.640329318]: Subscribed to topic:/camera/depth/camera_info [queue_size: 1][tcp_nodelay: 0]
Reading the mesh file original.stl
Info,  T0: Load 0p���.stl
Info,  T0: Found a matching importer for this file format
Info,  T0: Import root directory is './'
Info,  T0: Entering post processing pipeline
Info,  T0: Points: 0, Lines: 0, Triangles: 1, Polygons: 0 (Meshes, X = removed)
Error, T0: FindInvalidDataProcess fails on mesh normals: Found zero-length vector
Info,  T0: FindInvalidDataProcess finished. Found issues ...
Info,  T0: GenVertexNormalsProcess finished. Vertex normals have been calculated
Error, T0: Failed to compute tangents; need UV data in channel0
Info,  T0: JoinVerticesProcess finished | Verts in: 1536 out: 258 | ~83.2%
Info,  T0: Cache relevant are 1 meshes (512 faces). Average output ACMR is 0.669922
Info,  T0: Leaving post processing pipeline
Loaded 702d35bb8aa818369dc24ff51b000e44 with the number of samples 5100

OpenCV Error: Assertion failed (response_map.cols % T == 0) in linearize, file /tmp/binarydeb/ros-kinetic-opencv3-3.3.1/opencv_contrib/rgbd/src/linemod.cpp, line 1099
Traceback (most recent call last):
  File "/home/yuhan/ork_ws/src/object_recognition_core/apps/detection", line 24, in <module>
    run_plasm(args, plasm)
  File "/home/yuhan/ork_ws/src/ecto/python/ecto/opts.py", line 85, in run_plasm
    sched.execute(options.niter)
ecto.CellException:            exception_type  CellException
[cell_name] = pipeline1

[cell_type] = ecto_linemod::Detector

[function_name] = process_with_only_these_inputs

[type] = cv::Exception

[what] = /tmp/binarydeb/ros-kinetic-opencv3-3.3.1/opencv_contrib/rgbd/src/linemod.cpp:1099: error: (-215) response_map.cols % T == 0 in function linearize

I have no idea why it happened, it is a problem about resolution of figures or sth else?

thx.

I get a "freeglut illegal glutInit() reinitialization attempt" error when training the linemod detector. Looks like glutInit() is called when training for a object starts, with multiple objects it is then called multiple times --> error? I'm working with a self-compiled trunk version.

The full console output was:

Training 2 objects.
computing object_id: b1b5472a11f7085f8e9efc3b32026520
Info, T0: Load /tmp/file0Fnbws.stl
Info, T0: Found a matching importer for this file format
Info, T0: Import root directory is '/tmp/'
Info, T0: Entering post processing pipeline
Info, T0: Points: 0, Lines: 0, Triangles: 1, Polygons: 0 (Meshes, X = removed)
Error, T0: Failed to compute tangents; need UV data in channel0
Info, T0: JoinVerticesProcess finished | Verts in: 5601 out: 5599 | ~0.0%
Info, T0: Cache relevant are 1 meshes (1867 faces). Average output ACMR is 2.998929
Info, T0: Leaving post processing pipeline
computing object_id: b1b5472a11f7085f8e9efc3b32012fe6
freeglut illegal glutInit() reinitialization attempt

Hi Sir,

I follow the web site to enable object-recognition-kitchen.
I can setup the database successfully, and I also can see mesh via web_ui.

However, when I training it, the segmentation fault will happen,

I check the source I built, there is no error neither, thus, I have no idea about that

Can anyone give me some tips ? Thanks a lot.

This is call:

rosrun object_recognition_core training -c `rospack find object_recognition_linemod`/conf/training.ork

got this error:

Traceback (most recent call last):
  File "/home/destogl/Workspace/ros/catkin_ws/src/object_recognition_core/apps/training", line 19, in <module>
    pipeline_class = find_class([ pipeline_param['module'] ], pipeline_param['type'] )
  File "/home/destogl/Workspace/ros/catkin_ws/src/object_recognition_core/python/object_recognition_core/utils/find_classes.py", line 141, in find_class
    classes = find_cells(modules, base_types)
  File "/home/destogl/Workspace/ros/catkin_ws/src/object_recognition_core/python/object_recognition_core/utils/find_classes.py", line 95, in find_cells
    potential_cells = find_classes(modules, [])
  File "/home/destogl/Workspace/ros/catkin_ws/src/object_recognition_core/python/object_recognition_core/utils/find_classes.py", line 37, in find_classes
    m = __import__(module)
  File "/home/destogl/Workspace/ros/catkin_ws/devel/lib/python2.7/dist-packages/object_recognition_linemod/__init__.py", line 34, in <module>
    exec(__fh.read())
  File "<string>", line 1, in <module>
  File "/home/destogl/Workspace/ros/catkin_ws/src/linemod/python/object_recognition_linemod/detector.py", line 7, in <module>
    import ecto_cells.ecto_linemod as ecto_linemod
ImportError: /home/destogl/Workspace/ros/catkin_ws/devel/lib/libobject_recognition_renderer_glut.so: undefined symbol: glFramebufferTexture2D