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

what is the massive_epoch_240 checkpoint? #22

Closed
rishabh063 opened this issue May 4, 2024 · 3 comments
Closed

what is the massive_epoch_240 checkpoint? #22

rishabh063 opened this issue May 4, 2024 · 3 comments

Comments

@rishabh063
Copy link

hey saw i new checkpoint in stuff what is it?

@ZhengPeng7
Copy link
Owner

Aha, thx for the continual attention :)
I did a similar thing in InSPyReNet to train the model with massive datasets. In my setting, the massive training set is [DIS-TR, DIS-TEs] + [HRSOD-TR, HRSOD-TE] + [UHRSD-TR, UHRSD-TE] + [HRS10K-TR, HRS10K-TE], and the DIS-VD is left as validation set. You can find it in config.py.

@ZhengPeng7 ZhengPeng7 changed the title what is the massive_epoch_40 checkpoint? what is the massive_epoch_240 checkpoint? May 4, 2024
@rishabh063
Copy link
Author

rishabh063 commented May 4, 2024

Thanks for the training , how are the results when compared to closed source solutions ?

i made a quick colab file for people to try this repo in a easier way for single image inference ( i have written this 5-10 times but always loose it edit the same colab like a noob)

attaching the link : - https://colab.research.google.com/drive/1aCBF9THgyHHxJwoIYfkYfHOjv6m0y1A8?usp=sharing.

Do make a copy and add to read me ( if you like it ) . People new to ml (like me ) will like this kind of inference

@ZhengPeng7
Copy link
Owner

Good! Thanks. I'll add the link to it in the README.
About the model trained on massive data, it's not better than the BiRefNet-DIS on DIS-VD. But I think it should work better on the wild images.

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