Can you share the model weights tested on the BTCV dataset？

ljwztc / CLIP-Driven-Universal-Model

[ICCV 2023] CLIP-Driven Universal Model; Rank first in MSD Competition.

Other

576 stars 70 forks source link

Can you share the model weights tested on the BTCV dataset？ #5

Closed min121101 closed 1 year ago

min121101 commented 1 year ago

Hello！ Thank you for your incredible work! Can you share the model weights tested on the BTCV dataset？

ljwztc commented 1 year ago

Please stay tuned~ We would release the model weights later.

min121101 commented 1 year ago

Can you share the details of the dataset processing? I noticed that the official BTCV dataset has 83 items, but only 47 items in PAOT.txt.

ljwztc commented 1 year ago

Can you share the details of the dataset processing? I noticed that the official BTCV dataset has 83 items, but only 47 items in PAOT.txt.

As shown in official web Under Institutional Review Board (IRB) supervision, 50 abdomen CT scans of were randomly selected from a combination of an ongoing colorectal cancer chemotherapy trial, and a retrospective ventral hernia study, there are 50 CT scans and 3 of them are missing when we downloaded the data.

min121101 commented 1 year ago

Hi, we ran out of memory while trying to reproduce your model. Did you use eight NVIDIA RTX A5000 during training?

ljwztc commented 1 year ago

Hi, we ran out of memory while trying to reproduce your model. Did you use eight NVIDIA RTX A5000 during training?

Yes. You can decrease the --batch_size and --num_samples to save memory.

MrGiovanni commented 1 year ago

Dear @min121101

We just released the trained checkpoints of universal models. Please take a look at the README.

Best,

Zongwei