Skip to content
This repository has been archived by the owner on Nov 23, 2023. It is now read-only.

executing Execute the training file, but it has not been able to converge, and the prediction result of the verification data set is wrong. #4

Open
zxm115 opened this issue Sep 27, 2018 · 3 comments

Comments

@zxm115
Copy link

zxm115 commented Sep 27, 2018

 Excuse me , I want to ask you a question about executing the training file named train_aaf.py  using PASCAL VOC dataset, which it has not been able to converge, despite following the default parameter settings consistent with the paper. And the prediction result of the validation data set is wrong, sometimes it can display black images with noise points, but in most cases  it shows all black, I suspect that there is a problem with model training. However when I download the trained models which you offer, the results based on the test data set are consistent with the paper. Could you please give me some advice about this situation, thanks so much. If you could offer more detailed parameter setting instructions, I would be grateful.
@twke18
Copy link
Owner

twke18 commented Sep 27, 2018

Hi, please check our bashscripts examples. The parameters listed in the bash files are the ones we reported in the paper.

For training, perhaps you can first try training the baseline models for debugging.

@zxm115
Copy link
Author

zxm115 commented Sep 27, 2018

screenshot_2018-09-27 tensorboard 1
Thank you very much for your help. Can you help me see this image. Why is the prediction result of the validation data set above always showing no content?

@wanghaokevin
Copy link

screenshot_2018-09-27 tensorboard 1
Thank you very much for your help. Can you help me see this image. Why is the prediction result of the validation data set above always showing no content?

Have you solved this problem?I am training the baseline with the parameters of the bashscipts, but there is also no concept .

# for free to subscribe to this conversation on GitHub. Already have an account? #.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants