loss_fn.utils package

Submodules

loss_fn.utils.build_helper module

loss_fn.utils.build_helper.build_cls_teacher_from_opts(opts: Namespace) → Module

Helper function to build a classification teacher model from command-line arguments

Parameters:

opts – command-line arguments

Returns:

A teacher model

loss_fn.utils.class_weighting module

loss_fn.utils.class_weighting.compute_class_weights(target: Tensor, n_classes: int, norm_val: float = 1.1) → Tensor

Implementation of a class-weighting scheme, as defined in Section 5.2 of ENet paper.

Parameters:

• target – Tensor of shape [Batch_size, *] containing values in the range [0, C).

• n_classes – Integer specifying the number of classes \(C\)

• norm_val – Normalization value. Defaults to 1.1. This value is decided based on the

• <https (`ESPNetv2 paper) –

  //arxiv.org/abs/1811.11431>`_.

• Link – https://github.com/sacmehta/ESPNetv2/blob/b78e323039908f31347d8ca17f49d5502ef1a594/segmentation/loadData.py#L16

Returns:

A \(C\)-dimensional tensor containing class weights

Module contents