zzzqzhou / ram-dsir Goto Github PK
View Code? Open in Web Editor NEW[ECCV'22] Generalizable Medical Image Segmentation via Random Amplitude Mixup and Domain-Specific Image Restoration
[ECCV'22] Generalizable Medical Image Segmentation via Random Amplitude Mixup and Domain-Specific Image Restoration
Hello, the author. After checking your paper, we found that you are still experimenting on the prostate dataset. Can you open the download path of this dataset and run the code on this dataset? We look forward to your reply.
What's the environment of your implementation (including cuda and cudnn)? AND I don't know how to put the datasets into folder properly~ :-(
This problem occurred while training the prostate dataset. How to solve it? Looking forward your reply. Thanks in advance.
i use pytorch==1.8
Hi! I am currently working on replicating the experiments with the generalized model mentioned in the paper. While the model performs normally on other domains, I have observed a significant performance drop of approximately 10% compared to the reported baseline when training on five additional domains and testing on the BIDMC dataset.
To further investigate this issue, I would like to request access to the code files related to this specific experiment. This will help me understand the training and evaluation procedures used, as well as any potential hyperparameter settings that might be influencing the results.
I believe having access to these code files will assist me in diagnosing the cause of the performance drop and potentially suggesting improvements or alternative approaches. Thank you for considering my request and for your support in resolving this issue.
Looking forward to your response.
Best regards
Hi,
Thanks for your great work!
Are the reported results from the best epoch or the last epoch? Thank you.
Hello, the author. After downloading your public prostate dataset, it is found that it includes Domain1-Domain6, 'ISBI' and 'ISBI_ 1.5 ',' I2CVB ',' UCL ',' BIDMC ',' HK 'folders. After testing, I found that Domain1-Domain6 folders were used in training, but only' ISBI 'and' ISBI 'could be used in testing_ 1.5 ',' I2CVB ',' UCL ',' BIDMC ',' HK 'folders, but I am not clear about the Domain1-Domain6 folder and the' ISBI ',' ISBI_ Please inform us of the correspondence between the folders 1.5 ',' I2CVB ',' UCL ',' BIDMC 'and' HK 'and look forward to your reply.
hi, author,
I just want to run the code on prostate dataset and meet some mistakes. I use command as below:
cd code
python -W ignore train.py --data_root ../dataset --dataset prostate \
--domain_idxs 1,2,3,4,5 --test_domain_idx 0 \
--rec --is_out_domain --consistency --consistency_type kd \
--save_path ../outdir/prostate/target0 --gpu 0
Did I forget something?Looking forward your reply. Thanks in advance.
Hello author!
I found that the preprocessed prostate data downloaded from the link you provided has minimal value -1.0 and maximal value about 0.4 (all domains), but it mentioned that "We normalize the data individually to [-1, 1] in intensity values on both datasets" in this related paper. And I found there is no addtional data transformation method in the prostate dataloader. So why these two scale are different? And can you open the preprocessing code of prostate dataset? Thanks, thanks!!!
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.