SegLossBias icon indicating copy to clipboard operation
SegLossBias copied to clipboard

Published 20 hours ago • verified publisher icon by-liu
→

Metadata

Code for the paper : Do we really need dice? The hidden region-size biases of segmentation losses. MeDIA 2023. https://www.sciencedirect.com/science/article/abs/pii/S136184152300275X

Results 2 SegLossBias issues
Sort by recently updated
recently updated
newest added

Why are the tests/directories differently structured?

1 comment

`(base) :~/loss/SegLossBias/tests$ python test_compound_losses.py Traceback (most recent call last): File "~/SegLossBias/tests/test_compound_losses.py", line 5, in from seglossbias.modeling.compound_losses import CrossEntropyWithL1, CrossEntropyWithKL ModuleNotFoundError: No module named 'seglossbias' ` I simply cloned and tried...

aymuos15 avatar aymuos15

the testing accuracy is very low on retinal-lesions

```bash python tools/test_net.py --config-file ./configs/retinal-lesions/unet_bce-l1_896x896.yaml TEST.CHECKPOINT_PATH /home/acer/retinal-lesions_r50unet_896x896_bce-l1.pth ``` the results of DSC was only 0.0126. why?

twmht avatar twmht

← Metadata

32
Stars
3
Forks
Watchers

Owner

verified publisher icon by-liu

Metadata

Code for the paper : Do we really need dice? The hidden region-size biases of segmentation losses. MeDIA 2023. https://www.sciencedirect.com/science/article/abs/pii/S136184152300275X

Back