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

Regarding training , #1

Closed
narennadig opened this issue Nov 20, 2022 · 4 comments
Closed

Regarding training , #1

narennadig opened this issue Nov 20, 2022 · 4 comments

Comments

@narennadig
Copy link

Hey,
I replaced the images in the (input/tennis/frames/) folder with the ones I have, I have kept it for training, how long does it take?

It's been 1 hr with 346 epochs now

@guy-erez
Copy link
Collaborator

guy-erez commented Nov 20, 2022

Hello,
It depends on the length of the movie, and the graphics card you are using. In general, the learning can be slow , partially because the current code is not optimized. Inference, however, is fast.

@narennadig
Copy link
Author

Thanks!

One last question,

How many Epochs for STN and BMN do we ideally need to get good results, we have around 360 frames to train.

@guy-erez
Copy link
Collaborator

The amount of epochs needed depends (as in any learning task) on the complexity of the data . In this case it depends on complexity of the camera-movement. From our experiments, around ~5000 epochs for the STN and ~1000 epochs for the BMN should be fine.

@guy-erez guy-erez closed this as completed Dec 4, 2022
@guy-erez
Copy link
Collaborator

guy-erez commented Dec 4, 2022

I close this issue

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

2 participants