Description
🚀 Feature
A loss functions API in torchvision.
Motivation
The request is simple, we have loss functions available in torchvision
E.g. sigmoid_focal_loss , l1_loss. But these are quite scattered and we have to use torchvision.ops.sigmoid_focal_loss etc.
In future, we might need to include further loss functions. E.g. dice_loss
Since loss functions are differentiable we can put them under nn.
We can have
torchvision.nn.losses.sigmoid_focal_loss and so on.
This keeps the scope of nn open for other differentiable functions such as layers, etc.
Pitch
These losses are very specific and pertain to vision domain. These are really useful and in general not tied to any specific model.
Though the loss functions that we keep are usually in torch. If we keep under nn namespace, future migration stays simple.
instead of torchvision.nn.sigmoid_focal_loss it would be torch.nn.sigmoid_focal_loss.
This Pitch comes from the above issues.
More Loss Functions
Alternatives
Alternatively, this should go in torch. But if we keep the above idea, we can support them in torchvision and later deprecate and move to torch (when needed).
Currently, we include them under ops but it is actually not an operation it is a differentiable loss function.
Whereas other ops are not differentiable and perform transformations / some manipulation over boxes/layers.
Additional context
Here is a list of loss functions we would like to include.
- LabelSmoothing Loss ENH Adds label_smoothing to cross entropy loss pytorch#63122
- SoftTarget CrossEntropy Support for target with class probs in CrossEntropyLoss pytorch#61044
- Huber Loss Huber loss pytorch#50553
- Barron loss Implemented in classy vision
- JSD Loss
- Dice Loss
- Poly Loss
- gIoU Loss Used in DETR.
- Refactor Current Focal Loss from ops to nn.
- Refactor FRCNN Smooth L1 Loss to nn.
- Super Loss [Feature Request] SuperLoss (NeurIPS 2020) pytorch#49851
- TripletMarginLoss This has similar issue to
LabelSmoothing.TripletMarginLossis supported in PyTorch but we use a variant of it in torchvision references for similarity search. - DeepLabCELoss This is implemented in Detectron2, but in torchvision references and model training we use
nn.CrossEntropy()with a little modification to aux loss. - Multi Class Focal Loss
- PSNR Loss Also PSNR as
torchvision.opswill be nice. - Distance-IoU & Complete-IoU loss - see here - Added CIOU loss function #5776 Distance IoU #5786
- SioU loss
- Federated loss
- Poly Loss Adding Polyloss to
torchpytorch#76732 - Tversky Loss
References
We can refer to Kornia, Fvcore and few PyTorch issues that need this feature.