urbanlc.model.pipeline_transforms.mix_patch#

urbanlc.model.pipeline_transforms.mix_patch(image: Tensor, gt: Tensor, alpha: Optional[float] = 1.0, beta: Optional[float] = 1.0) → Tuple[Tensor, Tensor]#

Mix patches in the input image and ground truth following Beta distribution.

Parameters:

  • image (torch.Tensor) – Input image tensor.

  • gt (torch.Tensor) – Ground truth tensor.

  • alpha (Optional[float]) – Alpha parameter for beta distribution (default: 1.0).

  • beta (Optional[float]) – Beta parameter for beta distribution (default: 1.0).

Returns:

Mixed image and ground truth tensors.

Return type:

Tuple[torch.Tensor, torch.Tensor]