search

hnjzbss / EKAGen

[CVPR 2024]Instance-level Expert Knowledge and Aggregate Discriminative Attention for Radiology Report Generation
Apache License 2.0
15 stars 0 forks source link

datasets #2

Open banbooliang opened 1 month ago

banbooliang commented 1 month ago

How to generate the file "data/iu_xray/images300_array/"?

Eomtee commented 1 month ago

I think, run 'generate_adm.py' in ADM.

banbooliang commented 1 month ago

Thank you very much. However, I can't find 'MIMIC_best_weight.pth' in generate_adm.py and how can i get it?

Eomtee commented 1 month ago

you can download the weight from hugging face

dyy1201 commented 1 month ago

Could you please tell me which version of this mimic_cxr is? Does this version contain the images300 file directly?

hnjzbss commented 1 month ago

Thank you very much. However, I can't find 'MIMIC_best_weight.pth' in generate_adm.py and how can i get it?

You can download from https://huggingface.co/ShenshenBu/EKAGen, or train using the project at https://github.com/hnjzbss/medical-classification.

hnjzbss commented 1 month ago

Could you please tell me which version of this mimic_cxr is? Does this version contain the images300 file directly?

Images300 refers to downsampling the original images to 300×300 in order to reduce preprocessing time complexity.

dyy1201 commented 1 month ago

Could you please tell me which version of this mimic_cxr is? Does this version contain the images300 file directly?

Images300 refers to downsampling the original images to 300×300 in order to reduce preprocessing time complexity.

Thank you for the clarification regarding the Images300 downsampling. Could you please provide more details about the updated README .

banbooliang commented 1 month ago

Thanks for your reply. I want to use the iu_best_weight.pth to generate the IU_xray images200 files. Could you provide it? Thanks!

hnjzbss commented 1 month ago

Thank you very much. However, I can't find 'MIMIC_best_weight.pth' in generate_adm.py and how can i get it?

The IU X-ray dataset does not have multi-class labels, so for segmentation on IU X-ray, we used 'MIMIC_best_weight.pth' for inference.

hnjzbss commented 1 month ago

Could you please tell me which version of this mimic_cxr is? Does this version contain the images300 file directly?

Images300 refers to downsampling the original images to 300×300 in order to reduce preprocessing time complexity.

Thank you for the clarification regarding the Images300 downsampling. Could you please provide more details about the updated README .

Sure! Thank you for your interest in EKAGen. We will be updating the detailed README soon.