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

Continue Training possible? #13

Closed
wizpig opened this issue Aug 26, 2020 · 2 comments
Closed

Continue Training possible? #13

wizpig opened this issue Aug 26, 2020 · 2 comments

Comments

@wizpig
Copy link

wizpig commented Aug 26, 2020

Hi everyone,

very nice container setup you've done there! :)

After I finished a training run, I can't seem to find a way to continue the training, i.e. to start a new training with my pre-trained weights. I could probably create a hack by modifying the container, but I wonder if there is a way how you meant it to be.

Alex

@hadikoub
Copy link
Member

hadikoub commented Aug 26, 2020

Hello,

If you mean you want to use a weights file that u have already trained to do another training as a pre-trained weights. Yes, you can do that by placing the yolo_custom.weights file that you obtain in the first training in the dataset folder that you want to use and change the train_conifg.json model field to

"model": {
        "framework": "darknet",
        "model_name": "yolov4",
	    "custom_weights": {
            "enable": true,
            "name": "yolo_custom.weights"
        },

the dataset folder should look like :

/custom_dataset/:
                           /images
                           /labels
                           /train_config.json
                           /yolo_custom.weights

and then you can start the training

@wizpig
Copy link
Author

wizpig commented Sep 1, 2020

Sorry for the late reply. Thanks alot for your answer.:)

@wizpig wizpig closed this as completed Sep 1, 2020
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