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

About training code and data set #2

Open
l1haodong opened this issue Jul 30, 2020 · 5 comments
Open

About training code and data set #2

l1haodong opened this issue Jul 30, 2020 · 5 comments

Comments

@l1haodong
Copy link

Hi, nice work!

Sorry to bother you, if it's convenient, would you consider uploading training codes and datasets later?

Thank you sincerely!

@l1haodong l1haodong reopened this Jul 30, 2020
@JunHeum
Copy link
Owner

JunHeum commented Aug 25, 2020

Hi, l1haodong.

Sorry to late answer. we will upload train codes and datasets later.

If you want to use train codes as soon as possible, we recommend re-implement train codes.

In the paper, we describe the details of how to train.

Feel free to ask details of the training if you have any questions.

Thank you for your interest.

@972461099
Copy link

Hi, l1haodong.

Sorry to late answer. we will upload train codes and datasets later.

If you want to use train codes as soon as possible, we recommend re-implement train codes.

In the paper, we describe the details of how to train.

Feel free to ask details of the training if you have any questions.

Thank you for your interest.

Hi,I am try to re-implement train codes,but something I dont knew.
What is the training data set of Bilateral motion estimation and what is its label?

@JunHeum
Copy link
Owner

JunHeum commented Nov 17, 2020

Hi, 972461099.

  1. What is the training set of Bilateral motion estimation?
  • As mentioned in the paper, we use the 'train set of Vimeo90K' dataset to train the bilateral motion network.
  1. What is its label?
  • If what you said label was bilateral motion field ground truth, we use the intermediate frame of the triplet (consecutive three frames) as ground truth.

  • And we define the loss between the intermediate frame and the warped frame which is warped using predicted bilateral motion fields.

  • Finally, the bilateral motion network learn to estimate accurate bilateral motion fields which decrease this loss.

Thank you for your interest!

@fupiao1998
Copy link

Hello, do you have any plans to open source the training code? I look forward to your answer.

@Yuqin-gif
Copy link

Hello, will you consider uploading your training code?
Thank you sincerely!

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

5 participants