Hello,
I trained the model for 100K steps for the LiTS dataset (https://competitions.codalab.org/competitions/17094#learn_the_details), 3D CT images of liver cancer segmentation. I then successfully executed the segmentation_sample.py file. But, it looks like this segmentation_sample.py file saves a *.png file with a generated 2D axial slice image. But, to run the evaluation metrics such as IoU, we need the generated segmentation mask image. Please provide code to perform the latter.
Currently, when I run the segementation_env.py code on the ground truth segmented NII image and the generated 2D axial slice *.png image, I get error message of division by zero not possible. (command: python scripts/segmentation_env.py --inp_pth /MedSegDiffCT1/data/input --out_pth MedSegDiff/MedSegDiffCT1/data/output)
Attached is the generated 2D axial slice jpeg image.
Hello, I trained the model for 100K steps for the LiTS dataset (https://competitions.codalab.org/competitions/17094#learn_the_details), 3D CT images of liver cancer segmentation. I then successfully executed the segmentation_sample.py file. But, it looks like this segmentation_sample.py file saves a *.png file with a generated 2D axial slice image. But, to run the evaluation metrics such as IoU, we need the generated segmentation mask image. Please provide code to perform the latter.
Currently, when I run the segementation_env.py code on the ground truth segmented NII image and the generated 2D axial slice *.png image, I get error message of division by zero not possible. (command: python scripts/segmentation_env.py --inp_pth /MedSegDiffCT1/data/input --out_pth MedSegDiff/MedSegDiffCT1/data/output)
Attached is the generated 2D axial slice jpeg image.
thank you for your help.