Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Pose estimation Error #37

Open
srinimd2005 opened this issue Dec 22, 2016 · 13 comments
Open

Pose estimation Error #37

srinimd2005 opened this issue Dec 22, 2016 · 13 comments

Comments

@srinimd2005
Copy link

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.

@fivef
Copy link

fivef commented Apr 10, 2017

Perhaps I have the same problem:

ork_colorhist_cut

My objects were trained via ork_capture.
The origin of the stl is at the bottom not in the center. @nlyubova Should we change https://github.com/wg-perception/reconstruction/blob/master/apps/mesh_object to create stls with the origin centered?

@JimmyDaSilva
Copy link

A quick shift of the mesh in blender should be enough

@fivef
Copy link

fivef commented Apr 10, 2017

Yes shifting manually works but what do you think of modifying https://github.com/wg-perception/reconstruction/blob/master/apps/mesh_object to move the origin e.g. with a meshlab filter?

@nlyubova
Copy link
Contributor

nlyubova commented Apr 12, 2017

The origin of the mesh should be in the center. But this shift between the recognized object and the real one (as you have) can be also because of other parameters. For example, the frames used in the config file should be the one for the RGB camera and not for the depth one. Is it the case for you?

@fivef
Copy link

fivef commented Apr 12, 2017

So you say depth_frame_id: 'camera_depth_optical_frame' in my detection.ros.ork should be set to the RGB camera frame?

After shifting the origin to the center the alignment improves but still is not perfect. I will further try to improve this by tuning icp_dist_min and px_match_min.

Another question: Is it ok to use the topic camera/rgb/image_rect_color and camera/depth_registered/image_raw as my Xtion driver doesn't give me the image_raw topic. Does reconstruction and detection work with the raw data + the camera_info or is it ok to input the rectified image_rect_color?

@nlyubova
Copy link
Contributor

Try to replace the depth topic to the RGB topic and you will see if there is any difference.

For detection, you can use rectified image.

@fivef
Copy link

fivef commented Apr 18, 2017

Setting the depth_frame_id to camera_rgb_optical_frame and playing with the topics didn't help.

@MengDes
Copy link

MengDes commented May 11, 2019

Hi i think i got the same problem ,have you solve the question ?

@TONG369
Copy link

TONG369 commented Oct 15, 2019

Hi i think i got the same problem ,have you solve the question ?
Hello, has your problem been solved? I met the same problem as you. How can I solve it?

@MengDes
Copy link

MengDes commented Oct 15, 2019

Hi i think i got the same problem ,have you solve the question ?
Hello, has your problem been solved? I met the same problem as you. How can I solve it?

the mesh's origin that ork provide not in center, you can ajdust the center,or find some new mesh file , then train your own template

@TONG369
Copy link

TONG369 commented Oct 15, 2019 via email

@MengDes
Copy link

MengDes commented Oct 15, 2019

Thanks for your reply. Do you know which file was modified? At 2019-10-15 10:35:59, "MengDes" notifications@github.com wrote: Hi i think i got the same problem ,have you solve the question ? Hello, has your problem been solved? I met the same problem as you. How can I solve it? the mesh's origin that ork provide not in center, you can ajdust the center,or find some new mesh file , then train your own template — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

it 's been a long time ,i am quiet not sure,you can search on internet

@TONG369
Copy link

TONG369 commented Oct 15, 2019 via email

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

6 participants