-
Notifications
You must be signed in to change notification settings - Fork 65
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
Great job!When the dataset will be release? #2
Comments
Hi, BossZard Thank you for your recognition and attention :) The specific date has not been determined yet due to some issues with licenses and copyrights. What we can confirm is that we will release the validation set first to facilitate verification by everyone (also because the validation set is smaller and has fewer conflicts). We will announce the details as soon as the data release is ready. |
@blueyo0 Hello, would you like to firstly provide a list of publicly available datasets used in this work? We may deal with the license and copyright issues ourselves. Thanks! |
Hi, some important training datasets are listed here. We fine-tuned a new version of SAM-Med3D called SAM-Med3D-turbo on these 44 datasets to enhance the performance further (we employed our own data split to prevent data leakage, but you can adjust the split to suit your needs). We hope this list can be of assistance to you.
|
Hi, @blueyo0 I notice these datasets is always in SAM-Med3D/utils/data_paths.py. So, what's new in the new version of ckpt? In other words, what are the factors that lead to the huge performance improvement? Just training fine-tuning on fewer datasets to get better results? Or did you fine-tune it for more time, resulting in a noticeable improvement in results? Are there any specific considerations for fine-tuning the dataset selection? Thank your solid work! |
Great job!When the dataset will be release :)
The text was updated successfully, but these errors were encountered: