Check whether the `num_classes` in head matches the length of `CLASSSES` in `dataset`. Args: runner (obj:`EpochBasedRunne

Get the expansion of a residual block. The block expansion will be obtained by the following order: 1. If ``expansion`` is given, just r

Get the expansion of a residual block. The block expansion will be obtained by the following order: 1. If ``expansion`` is given, just r

Get the expansion of a residual block. The block expansion will be obtained by the following order: 1. If ``expansion`` is given, just r

Paste instance masks acoording to boxes. This implementation is modified from https://github.com/facebookresearch/detectron2/ Args:

involution kernel

involution kernel

involution kernel

Transform outputs for a single batch item into bbox predictions. Args: cls_scores (list[Tensor]): Box scores for each scale level

Initialize layers of the head.

Check whether the training dataset is compatible with head. Args: runner (obj:`EpochBasedRunner`): Epoch based Runner.

Check whether the dataset in val epoch is compatible with head. Args: runner (obj:`EpochBasedRunner`): Epoch based Runner.

Forward computation Args: x (tensor | tuple[tensor]): x could be a Torch.tensor or a tuple of Torch.tensor, conta

Forward function.

Forward computation Args: x (tensor | tuple[tensor]): x could be a Torch.tensor or a tuple of Torch.tensor, conta

Forward feature map of a single scale level.

Only keep loading image and annotations related configuration. Args: pipeline (list[dict]): Data pipeline configs. Returns:

Get segmentation masks from mask_pred and bboxes. Args: mask_pred (Tensor or ndarray): shape (n, #class, h, w). F

Init backbone weights Args: pretrained (str | None): If pretrained is a string, then it initializes backbone weig

Initialize the weights of FPN module.

Init backbone weights Args: pretrained (str | None): If pretrained is a string, then it initializes backbone weig

Initialize weights of the head.

Compute losses of the head. Args: cls_scores (list[Tensor]): Box scores for each scale level Has shape (N, num_an