Segmentation masks have a different shape than the input CT scan?

ljwztc / CLIP-Driven-Universal-Model

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

Other

577 stars 70 forks source link

Segmentation masks have a different shape than the input CT scan? #81

Open chongduan opened 3 months ago

chongduan commented 3 months ago

Hi, I am using the pred_pseudo.py function to process a CT scan (from one of the Decathlon datasets), and I noticed that the output masks have a different shape than the input CT scan. I imagine this must be due to some transforms in the process, but I could not figure out which step is causing this shape inconsistency. Any help will be greatly appreciated!

ljwztc commented 3 months ago

Invertd() API in monai is used for save result. If you rigorously follow the pipeline in README, it should generate shape with same shape as input.