Giters

RobotLocomotion / pytorch-dense-correspondence

Code for "Dense Object Nets: Learning Dense Visual Object Descriptors By and For Robotic Manipulation"

Home Page:https://arxiv.org/pdf/1806.08756.pdf

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Would you please provide a pretrained model?

Wegnery219 opened this issue · comments

commented

It's so complex for me to get the data.Would you please provide a pretrained model?I want to see the project effect and try the evaluate function.

commented

Yes we will do this eventually, along with some sample data so you can evaluate the network.

commented

Thank you for your reply.

commented

Hi @Wegnery219, @JiamingSuen --
We have now made all of our data available!
This tutorial will walk through downloading the data and training a model.
If you are familiar with this type of software and have a new-ish GPU, I estimate only would take ~30 minutes to get a model trained and results visualized.
Let me know if you get caught up, or we could make anything more clear?

commented

Hi @Wegnery219, @JiamingSuen -- We have now made all of our data available! This tutorial will walk through downloading the data and training a model. If you are familiar with this type of software and have a new-ish GPU, I estimate only would take ~30 minutes to get a model trained and results visualized. Let me know if you get caught up, or we could make anything more clear?

Thank you for your work and open-sourcing it. Would you mind sharing a pretrained model @manuelli @peteflorence ? I try to re-create your Self-Supervised Correspondence in Visuomotor Policy Learning paper, which builds on the pre-trained model.

Nvm. I found the model zoo now.