a question about PoseNet !

146 views
Skip to first unread message

jaba marwen

unread,
Jun 23, 2017, 11:15:41 AM6/23/17
to Caffe Users

Hi every one ,

I want to reproduce the results of the paper PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization : Alex Kendall, Matthew Grimes and Roberto Cipolla .
But , I have a dumb question should I install their modified version of Caffe  caffe PoseNet or can I experiment on Caffe of BVLC repository .




Thanks

Przemek D

unread,
Jun 27, 2017, 2:14:09 AM6/27/17
to Caffe Users
The authors introduce some important changes to the repository, particularly to the data layer. If you want to reproduce their results, I'd say use their repo.

Jonathan R. Williford

unread,
Jun 27, 2017, 4:39:33 PM6/27/17
to Przemek D, Caffe Users
Yeah, this is the standard way of doing things until something like this gets merged:

Cheers,
Jonathan

--
You received this message because you are subscribed to the Google Groups "Caffe Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caffe-users+unsubscribe@googlegroups.com.
To post to this group, send email to caffe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/caffe-users/80471795-28ff-4769-82ff-8016ad083661%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

jaba marwen

unread,
Jun 29, 2017, 5:27:05 AM6/29/17
to Jonathan R. Williford, Przemek D, Caffe Users
Thank you for your answers , I managed to install the modified version of Caffe and reproduce the results . I just followed the documentation for installation installation

--
You received this message because you are subscribed to a topic in the Google Groups "Caffe Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/caffe-users/N5ZkdoMXiJQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to caffe-users+unsubscribe@googlegroups.com.

To post to this group, send email to caffe...@googlegroups.com.

ming.he...@gmail.com

unread,
Apr 20, 2018, 6:45:53 AM4/20/18
to Caffe Users
Hi Jaba marwen,

Could you pls advise me that were you be able to reproduce the result of PoseNet over the Street scene. The translation error of my model is around 25m and seems never gonna converge to ~3m in the paper.
Cheers,

On Thursday, June 29, 2017 at 6:57:05 PM UTC+9:30, jaba marwen wrote:
Thank you for your answers , I managed to install the modified version of Caffe and reproduce the results . I just followed the documentation for installation installation
On Tue, Jun 27, 2017 at 10:39 PM, Jonathan R. Williford <jonathan.r...@gmail.com> wrote:
Yeah, this is the standard way of doing things until something like this gets merged:

Cheers,
Jonathan
On Tue, Jun 27, 2017 at 8:14 AM, Przemek D <snowb...@gmail.com> wrote:
The authors introduce some important changes to the repository, particularly to the data layer. If you want to reproduce their results, I'd say use their repo.

W dniu piątek, 23 czerwca 2017 17:15:41 UTC+2 użytkownik jaba marwen napisał:

Hi every one ,

I want to reproduce the results of the paper PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization : Alex Kendall, Matthew Grimes and Roberto Cipolla .
But , I have a dumb question should I install their modified version of Caffe  caffe PoseNet or can I experiment on Caffe of BVLC repository .




Thanks

--
You received this message because you are subscribed to the Google Groups "Caffe Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caffe-users...@googlegroups.com.

To post to this group, send email to caffe...@googlegroups.com.

--
You received this message because you are subscribed to a topic in the Google Groups "Caffe Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/caffe-users/N5ZkdoMXiJQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to caffe-users...@googlegroups.com.

To post to this group, send email to caffe...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages