pynif3d.loss¶
- pynif3d.loss.eikonal_loss(x)¶
Computes the eikonal loss for a given set of points.
- Parameters
x (torch.Tensor) – Tensor containing the point coordinates. Its shape is
(batch_size, n_samples, 3)
or(n_samples, 3)
.- Returns
Tensor containing the eikonal loss. Its shape is
(1,)
.- Return type
torch.Tensor
- pynif3d.loss.mse_to_psnr(mse_loss, max_intensity=1.0)¶
Converts a mean-squared error (MSE) loss to peak signal-to-noise ratio (PSNR).
- Parameters
mse_loss (torch.Tensor) – MSE loss. Its shape is
(1,)
.max_intensity (float) – The maximum pixel intensity. Default value is 1.0.
- Returns
Tensor with the corresponding PSNR loss.
- Return type
torch.Tensor