Best Way to Reweight Faster R-CNN Loss?

I’m trying to figure out what’s the best way to reweight the component losses in Faster R-CNN.

As far as I can tell this part of the code returns the component loss (vision/ at af97ec2f4c9daac091b9a87355c4f22d37488004 · pytorch/vision · GitHub):

def fastrcnn_loss(class_logits, box_regression, labels, regression_targets):
    # type: (Tensor, Tensor, List[Tensor], List[Tensor]) -> Tuple[Tensor, Tensor]
    Computes the loss for Faster R-CNN.
        class_logits (Tensor)
        box_regression (Tensor)
        labels (list[BoxList])
        regression_targets (Tensor)
        classification_loss (Tensor)
        box_loss (Tensor)

    labels =, dim=0)
    regression_targets =, dim=0)

    classification_loss = F.cross_entropy(class_logits, labels)

    # get indices that correspond to the regression targets for
    # the corresponding ground truth labels, to be used with
    # advanced indexing
    sampled_pos_inds_subset = torch.where(labels > 0)[0]
    labels_pos = labels[sampled_pos_inds_subset]
    N, num_classes = class_logits.shape
    box_regression = box_regression.reshape(N, box_regression.size(-1) // 4, 4)

    box_loss = det_utils.smooth_l1_loss(
        box_regression[sampled_pos_inds_subset, labels_pos],
        beta=1 / 9,
    box_loss = box_loss / labels.numel()

    return classification_loss, box_loss

I can add some custom weights to change the importance of classification vs box loss.

What’s the best way of modify this function, but without needing to create a separate ecosystem of custom FasterRCNN classes and objects?