-
Hello,
I'm trying to use `ArrayDataset` to train a UNet with a single channeled image. I have my own data augmentation procedure that passes array data of size (160, 160) to `ArrayDataset`. I use t…
-
The Retina UNet's seg dice loss never goes down while training
![image](https://user-images.githubusercontent.com/21152737/49907733-7972ca80-feb2-11e8-8974-56be411f9147.png)
And the validation seg…
-
Hi @pfjaeger , thank you for sharing this great repository! My pred_example_{}.png (for Mask R-CNN) is returning additional coloured boxes that are purple and white, see below:
![image](https://use…
-
1. Need to update Readme on the calibration dataset specifics.
--use first 128k pairs in the second half of day 23. Test set is first half of day 23.
2. Need a standalone script to calculate acc…
-
Traceback (most recent call last):
File "Example_on_BRATS2018.py", line 109, in
model.fit(data, labels, batch_size=1, epochs=1)
File "/home/xxx/anaconda3/envs/tf108/lib/python3.6/site-pack…
-
They do this in MITK-Diffusion [[github link]](https://github.com/MIC-DKFZ/MITK-Diffusion)
Look for "BetData". They basically add it as a dependency (like GuidelinesSupportLibrary) in ```CMake/``` …
dboun updated
4 years ago
-
**Describe the bug**
It says error: "No valid images selected!"
**To Reproduce**
Steps to reproduce the behavior:
1. Load 1+ (brain?) images
2. Load their segmentation
3. feature extraction pa…
dboun updated
4 years ago
-
-
As the image downscales x2 with each step along each dimension, it is impossible for it to deal with images that are uneven in any dimension. In decathlon dataset (and brats as well), the number of sl…
-
Iam trying this approach on 2019 brats dataset with same model. The VAE dice coefficient (reconstructed image ) is around 0.90 but the dice coefficient for the segmentation label is 0.0000e+00.
…