GaryGuTC / COMG_model

[WACV 2024] Complex Organ Mask Guided Radiology Report Generation
MIT License
30 stars 2 forks source link

MIMIC-CXR #12

Open ecust-sl opened 4 months ago

ecust-sl commented 4 months ago

Hello, I would like to ask if you still have the pre-processed files for the MIMIC-CXR dataset can you share them, the speed of processing them on my side seems to be very slow, it takes many days I don't know what's going on?

GaryGuTC commented 4 months ago

Hi  The pre-processed files are too vast and hard to upload to any drive. The image number in the MIMIC-CXR dataset is too huge; it needs more time to process, almost 5h using a single 3090. Please check whether you use GPU to process it.

Best regards

ecust-sl commented 4 months ago
image image image

I'm using a single 4090 card. Why is it taking so long?

ecust-sl commented 4 months ago

Hi  The pre-processed files are too vast and hard to upload to any drive. The image number in the MIMIC-CXR dataset is too huge; it needs more time to process, almost 5h using a single 3090. Please check whether you use GPU to process it.

Best regards

image image image I'm using a single 4090 card. Why is it taking so long?

Hi  The pre-processed files are too vast and hard to upload to any drive. The image number in the MIMIC-CXR dataset is too huge; it needs more time to process, almost 5h using a single 3090. Please check whether you use GPU to process it.

Best regards

I'm using a single 4090 card, why is it so slow, it doesn't make sense, I downloaded the model locally myself, does it have anything to do with this?

GaryGuTC commented 4 months ago

Hi 4090 is faster than 3090 in my mind.

  1. (important!) Please check whether you use GPU to do the inference. (65w/450w may not use GPU)
  2. Please check that the MIMIC-CXR dataset used is downloaded from here Which has been pre-processed.

Hope that can help you. Best regard,

ecust-sl commented 4 months ago
image

Hi 4090 is faster than 3090 in my mind.

  1. (important!) Please check whether you use GPU to do the inference. (65w/450w may not use GPU)
  2. Please check that the MIMIC-CXR dataset used is downloaded from here Which has been pre-processed.

Hope that can help you. Best regard,

image

The gpu is indeed in use, the video memory usage is about these, the dataset was downloaded from the official website

ecust-sl commented 4 months ago

Hi 4090 is faster than 3090 in my mind.

  1. (important!) Please check whether you use GPU to do the inference. (65w/450w may not use GPU)
  2. Please check that the MIMIC-CXR dataset used is downloaded from here Which has been pre-processed.

Hope that can help you. Best regard,

Hello, may I ask if the image format you are using is jpg or dcm?

GaryGuTC commented 4 months ago

Hi  Please try to use the dataset after processing

Best regards,

ecust-sl commented 4 months ago

Hi  Please try to use the dataset after processing

Best regards,

Hi

image

When I try to open this link, it says that I don't have enough permissions, the author of this hard disk has set the permissions, can you tell me about it?

GaryGuTC commented 4 months ago

Hi Please send email to apply.

Best regards,

anti-machinee commented 1 month ago

@GaryGuTC Hi I just send the request access to access MIMIC via https://github.com/zhjohnchan/R2GenCMN#datasets

Please can I have the access ?

GaryGuTC commented 1 month ago

Hi

This google drive file is not mine. Please send an email again to apply for permission.

Thanks a lot for your attention. Best regards,