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

How to continue training using last saved weights? #176

Open
MuhammadAsadJaved opened this issue Nov 14, 2019 · 1 comment
Open

How to continue training using last saved weights? #176

MuhammadAsadJaved opened this issue Nov 14, 2019 · 1 comment

Comments

@MuhammadAsadJaved
Copy link

MuhammadAsadJaved commented Nov 14, 2019

Hi ,
I have trained a pix2pix model on custom dataset for 200 epochs and tested it but the generated images are not very good. can you give me some suggestion to increase the performance?
1-I have trained model on 512 x 512 visual images to generate infrared images using paired images.
2- How can I continue the training from last stored weights? Do you think the performance will increase if i train this model more ? i.e for 1000 epoch.
3-Please see attached input and output examples generated with trained model.
3- Any other suggestions to enhance the quality of output?

1
1output
2
2output
3
3output
4
4output

@MuhammadAsadJaved
Copy link
Author

There is an argument in the pix2pix.py "--checkpoint". we can resume the training from the last saved weights using it. For example.

python3 pix2pix.py --mode train --output_dir FlirKaist_train --checkpoint FlirKaist_train --max_epochs 1000 --input_dir Data/combined/train --which_direction BtoA

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

1 participant