Module: sarpyx.utils.sar_loss
File: sarpyx/utils/sar_loss.py
Loss Functions for Complex-Valued SAR Image Super-Resolution. This module provides specialized loss functions for Synthetic Aperture Radar (SAR) image super-resolution tasks, handling both amplitude and phase components of complex-valued SAR data.
Exported Symbols (__all__)
No explicit __all__ list. Public symbols inferred from implementation.
Public Functions (15)
amplitude_l1_loss function
Compute L1 loss between magnitudes of predicted and target complex SAR images.
File location: sarpyx/utils/sar_loss.py:18
Signature
amplitude_l1_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor of shape (..., H, W) or (..., H, W, 2) |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the L1 amplitude loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import amplitude_l1_loss
result = amplitude_l1_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
amplitude_l2_loss function
Compute L2 loss (MSE) between magnitudes of predicted and target SAR images.
File location: sarpyx/utils/sar_loss.py:48
Signature
amplitude_l2_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor of shape (..., H, W) or (..., H, W, 2) |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the L2 amplitude loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import amplitude_l2_loss
result = amplitude_l2_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
amplitude_ssim_loss function
Compute SSIM loss between magnitudes of predicted and target SAR images.
File location: sarpyx/utils/sar_loss.py:79
Signature
amplitude_ssim_loss(pred: torch.Tensor, target: torch.Tensor, window_size: int=11) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor of shape (B, C, H, W) or (H, W) |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
window_size | int | no | 11 | Size of Gaussian window for SSIM computation |
Return Type
torch.Tensor
Scalar tensor representing the SSIM loss (1 - SSIM)
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import amplitude_ssim_loss
result = amplitude_ssim_loss(pred=<pred>, target=<target>)Edge Cases
Includes optional parameters with implementation-defined fallback behavior.
amplitude_log_cosh_loss function
Compute log-cosh loss between magnitudes of predicted and target SAR images.
File location: sarpyx/utils/sar_loss.py:151
Signature
amplitude_log_cosh_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor of shape (..., H, W) or (..., H, W, 2) |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the log-cosh amplitude loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import amplitude_log_cosh_loss
result = amplitude_log_cosh_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
phase_l1_loss function
Compute L1 loss between phases of predicted and target complex SAR images.
File location: sarpyx/utils/sar_loss.py:187
Signature
phase_l1_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the phase L1 loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import phase_l1_loss
result = phase_l1_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
phase_l2_loss function
Compute L2 loss between phases of predicted and target complex SAR images.
File location: sarpyx/utils/sar_loss.py:223
Signature
phase_l2_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the phase L2 loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import phase_l2_loss
result = phase_l2_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
phase_wrapping_error_loss function
Compute phase wrapping error loss between predicted and target SAR images.
File location: sarpyx/utils/sar_loss.py:257
Signature
phase_wrapping_error_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the phase wrapping error loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import phase_wrapping_error_loss
result = phase_wrapping_error_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
phase_correlation_loss function
Compute phase correlation loss between predicted and target complex SAR images.
File location: sarpyx/utils/sar_loss.py:283
Signature
phase_correlation_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the phase correlation loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import phase_correlation_loss
result = phase_correlation_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
complex_l2_loss function
Compute MSE between predicted and target complex SAR images.
File location: sarpyx/utils/sar_loss.py:325
Signature
complex_l2_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the complex L2 loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import complex_l2_loss
result = complex_l2_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
complex_correlation_loss function
Compute loss based on complex correlation between predicted and target SAR images.
File location: sarpyx/utils/sar_loss.py:366
Signature
complex_correlation_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the complex correlation loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import complex_correlation_loss
result = complex_correlation_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
coherence_loss function
Compute interferometric coherence loss between predicted and target SAR images.
File location: sarpyx/utils/sar_loss.py:404
Signature
coherence_loss(pred: torch.Tensor, target: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor |
target | torch.Tensor | yes | - | Target complex SAR image tensor of same shape as pred |
Return Type
torch.Tensor
Scalar tensor representing the coherence loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import coherence_loss
result = coherence_loss(pred=<pred>, target=<target>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
gan_generator_loss function
Compute adversarial loss for generator in GAN training.
File location: sarpyx/utils/sar_loss.py:444
Signature
gan_generator_loss(disc_pred: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
disc_pred | torch.Tensor | yes | - | Discriminator's output probabilities for generated images |
Return Type
torch.Tensor
Scalar tensor representing the generator's adversarial loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import gan_generator_loss
result = gan_generator_loss(disc_pred=<disc_pred>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
gan_discriminator_loss function
Compute discriminator loss in GAN for SAR super-resolution.
File location: sarpyx/utils/sar_loss.py:470
Signature
gan_discriminator_loss(disc_real: torch.Tensor, disc_fake: torch.Tensor) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
disc_real | torch.Tensor | yes | - | Discriminator predictions on real images |
disc_fake | torch.Tensor | yes | - | Discriminator predictions on generated images |
Return Type
torch.Tensor
Scalar tensor representing the discriminator loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import gan_discriminator_loss
result = gan_discriminator_loss(disc_real=<disc_real>, disc_fake=<disc_fake>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
perceptual_loss function
Compute perceptual feature loss using pretrained feature extractor.
File location: sarpyx/utils/sar_loss.py:502
Signature
perceptual_loss(pred: torch.Tensor, target: torch.Tensor, feature_extractor: nn.Module) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted SAR image tensor |
target | torch.Tensor | yes | - | Target SAR image tensor of same shape as pred |
feature_extractor | nn.Module | yes | - | Pretrained CNN for feature extraction (e.g., VGG) |
Return Type
torch.Tensor
Scalar tensor representing the perceptual loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import perceptual_loss
result = perceptual_loss(pred=<pred>, target=<target>, feature_extractor=<feature_extractor>)Edge Cases
No explicit edge-case section found; behavior is inferred from implementation.
combined_sar_loss function
Compute combined SAR loss with multiple components.
File location: sarpyx/utils/sar_loss.py:547
Signature
combined_sar_loss(pred: torch.Tensor, target: torch.Tensor, amplitude_weight: float=1.0, phase_weight: float=0.5, complex_weight: float=0.3, ssim_weight: float=0.2) -> torch.TensorParameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
pred | torch.Tensor | yes | - | Predicted complex SAR image tensor |
target | torch.Tensor | yes | - | Target complex SAR image tensor |
amplitude_weight | float | no | 1.0 | Weight for amplitude L1 loss |
phase_weight | float | no | 0.5 | Weight for phase L1 loss |
complex_weight | float | no | 0.3 | Weight for complex L2 loss |
ssim_weight | float | no | 0.2 | Weight for SSIM loss |
Return Type
torch.Tensor
Scalar tensor representing the combined loss
Exceptions
None explicitly documented; inferred from implementation.
Side Effects
- inferred from implementation
Example Usage
from sarpyx.utils.sar_loss import combined_sar_loss
result = combined_sar_loss(pred=<pred>, target=<target>)Edge Cases
Includes optional parameters with implementation-defined fallback behavior.
Public Classes (0)
No public classes detected.