trainers.segmentation JaccardLoss receiving num_classes, should be a List[int]? about torchgeo HOT 8 CLOSED

I verified the behavior of JaccardLoss here https://gist.github.com/calebrob6/658edaa59c68f0c0a510f8d9d7a41458.

Currently our code will always use the average jaccard over all classes regardless of what self.hparams["num_classes"] is.

We definitely should fix this.

@isaaccorley ignore_index support might be a cool feature for the torchseg fork.

from torchgeo.

TorchSeg now supports both classes and ignore_index in DiceLoss, JaccardLoss, and FocalLoss.

from torchgeo.

+1 for fiddling with weights in CrossEntropyLoss.

Yep, I can imagine a few scenarios too (e.g. dropping loss over noisy classes). I would usually do this by simply remapping all classes I didn't care about to a nodata mask value, then setting ignore_index to that value, but this is more elegant. My concern now would be implementing this functionality for the other loss functions.

from torchgeo.

@robmarkcole it's merged into main now. I'll make another pre-release on pypi today.

from torchgeo.

I suggest this requires a new arg of classes_in_loss which is a List[int]. This should pop off the ignore_index

from torchgeo.

Have you run into scenarios in which you would want to ignore multiple classes? If not, I think we can just pop ignore_index from the entire class list if it is given.

from torchgeo.

I've not really used JaccardLoss, and commonly use nn.CrossEntropyLoss where the weights parameter has been very effective for imbalanced datasets. If I understand the use of classes in JaccardLoss it has a simiar purpose, but is binary on/off per class, rather than a weighting

Coming back to your question, for multiclass datasets where I might only care about 1 or 2 classes, I can imagine ignoring multiple classes

from torchgeo.

Thanks @isaaccorley

For anyone looking, they are on branch https://github.com/isaaccorley/torchseg/blob/losses/ignore-index/torchseg/losses/

from torchgeo.