wavedl 1.6.0__py3-none-any.whl → 1.6.1__py3-none-any.whl

wavedl/models/mobilenetv3.py

@@ -1,295 +1,295 @@
-"""
-MobileNetV3: Efficient Networks for Edge Deployment
-====================================================
-
-Lightweight architecture optimized for mobile and embedded devices.
-MobileNetV3 combines neural architecture search (NAS) with hardware-aware
-optimization to achieve excellent accuracy with minimal computational cost.
-
-**Key Features**:
-    - Inverted residuals with depthwise separable convolutions
-    - Squeeze-and-Excitation (SE) attention for channel weighting
-    - h-swish activation: efficient approximation of swish
-    - Designed for real-time inference on CPUs and edge devices
-
-**Variants**:
-    - mobilenet_v3_small: Ultra-lightweight (~0.9M backbone params) - Edge/embedded
-    - mobilenet_v3_large: Balanced (~3.0M backbone params) - Mobile deployment
-
-**Use Cases**:
-    - Real-time structural health monitoring on embedded systems
-    - Field inspection with portable devices
-    - When model size and inference speed are critical
-
-**Note**: MobileNetV3 is 2D-only. For 1D data, use TCN. For 3D data, use ResNet3D.
-
-References:
-    Howard, A., et al. (2019). Searching for MobileNetV3.
-    ICCV 2019. https://arxiv.org/abs/1905.02244
-
-Author: Ductho Le (ductho.le@outlook.com)
-"""
-
-from typing import Any
-
-import torch
-import torch.nn as nn
-
-
-try:
-    from torchvision.models import (
-        MobileNet_V3_Large_Weights,
-        MobileNet_V3_Small_Weights,
-        mobilenet_v3_large,
-        mobilenet_v3_small,
-    )
-
-    MOBILENETV3_AVAILABLE = True
-except ImportError:
-    MOBILENETV3_AVAILABLE = False
-
-from wavedl.models.base import BaseModel
-from wavedl.models.registry import register_model
-
-
-class MobileNetV3Base(BaseModel):
-    """
-    Base MobileNetV3 class for regression tasks.
-
-    Wraps torchvision MobileNetV3 with:
-    - Optional pretrained weights (ImageNet-1K)
-    - Automatic input channel adaptation (grayscale → 3ch)
-    - Lightweight regression head (maintains efficiency)
-
-    MobileNetV3 is ideal for:
-    - Edge deployment (Raspberry Pi, Jetson, mobile)
-    - Real-time inference requirements
-    - Memory-constrained environments
-    - Quick prototyping and experimentation
-
-    Note: This is 2D-only. Input shape must be (H, W).
-    """
-
-    def __init__(
-        self,
-        in_shape: tuple[int, int],
-        out_size: int,
-        model_fn,
-        weights_class,
-        pretrained: bool = True,
-        dropout_rate: float = 0.2,
-        freeze_backbone: bool = False,
-        regression_hidden: int = 256,
-        **kwargs,
-    ):
-        """
-        Initialize MobileNetV3 for regression.
-
-        Args:
-            in_shape: (H, W) input image dimensions
-            out_size: Number of regression output targets
-            model_fn: torchvision model constructor
-            weights_class: Pretrained weights enum class
-            pretrained: Use ImageNet pretrained weights (default: True)
-            dropout_rate: Dropout rate in regression head (default: 0.2)
-            freeze_backbone: Freeze backbone for fine-tuning (default: False)
-            regression_hidden: Hidden units in regression head (default: 256)
-        """
-        super().__init__(in_shape, out_size)
-
-        if not MOBILENETV3_AVAILABLE:
-            raise ImportError(
-                "torchvision is required for MobileNetV3. "
-                "Install with: pip install torchvision"
-            )
-
-        if len(in_shape) != 2:
-            raise ValueError(
-                f"MobileNetV3 requires 2D input (H, W), got {len(in_shape)}D. "
-                "For 1D data, use TCN. For 3D data, use ResNet3D."
-            )
-
-        self.pretrained = pretrained
-        self.dropout_rate = dropout_rate
-        self.freeze_backbone = freeze_backbone
-        self.regression_hidden = regression_hidden
-
-        # Load pretrained backbone
-        weights = weights_class.IMAGENET1K_V1 if pretrained else None
-        self.backbone = model_fn(weights=weights)
-
-        # MobileNetV3 classifier structure:
-        # classifier[0]: Linear (features → 1280 for Large, 1024 for Small)
-        # classifier[1]: Hardswish
-        # classifier[2]: Dropout
-        # classifier[3]: Linear (1280/1024 → num_classes)
-
-        # Get the input features to the final classifier
-        in_features = self.backbone.classifier[0].in_features
-
-        # Replace classifier with lightweight regression head
-        # Keep it efficient to maintain MobileNet's speed advantage
-        self.backbone.classifier = nn.Sequential(
-            nn.Linear(in_features, regression_hidden),
-            nn.Hardswish(inplace=True),  # Match MobileNetV3's activation
-            nn.Dropout(dropout_rate),
-            nn.Linear(regression_hidden, out_size),
-        )
-
-        # Adapt first conv for single-channel input (3× memory savings vs expand)
-        self._adapt_input_channels()
-
-        # Optionally freeze backbone for fine-tuning (after adaptation so new conv is frozen too)
-        if freeze_backbone:
-            self._freeze_backbone()
-
-    def _adapt_input_channels(self):
-        """Modify first conv to accept single-channel input.
-
-        Instead of expanding 1→3 channels in forward (which triples memory),
-        we replace the first conv layer with a 1-channel version and initialize
-        weights as the mean of the pretrained RGB filters.
-        """
-        old_conv = self.backbone.features[0][0]
-        new_conv = nn.Conv2d(
-            1,  # Single channel input
-            old_conv.out_channels,
-            kernel_size=old_conv.kernel_size,
-            stride=old_conv.stride,
-            padding=old_conv.padding,
-            dilation=old_conv.dilation,
-            groups=old_conv.groups,
-            padding_mode=old_conv.padding_mode,
-            bias=old_conv.bias is not None,
-        )
-        if self.pretrained:
-            with torch.no_grad():
-                new_conv.weight.copy_(old_conv.weight.mean(dim=1, keepdim=True))
-        self.backbone.features[0][0] = new_conv
-
-    def _freeze_backbone(self):
-        """Freeze all backbone parameters except the classifier."""
-        for name, param in self.backbone.named_parameters():
-            if "classifier" not in name:
-                param.requires_grad = False
-
-    def forward(self, x: torch.Tensor) -> torch.Tensor:
-        """
-        Forward pass.
-
-        Args:
-            x: Input tensor of shape (B, 1, H, W)
-
-        Returns:
-            Output tensor of shape (B, out_size)
-        """
-        return self.backbone(x)
-
-    @classmethod
-    def get_default_config(cls) -> dict[str, Any]:
-        """Return default configuration for MobileNetV3."""
-        return {
-            "pretrained": True,
-            "dropout_rate": 0.2,
-            "freeze_backbone": False,
-            "regression_hidden": 256,
-        }
-
-
-# =============================================================================
-# REGISTERED MODEL VARIANTS
-# =============================================================================
-
-
-@register_model("mobilenet_v3_small")
-class MobileNetV3Small(MobileNetV3Base):
-    """
-    MobileNetV3-Small: Ultra-lightweight for edge deployment.
-
-    ~0.9M backbone parameters. Designed for the most constrained environments.
-    Achieves ~67% ImageNet accuracy with minimal compute.
-
-    Recommended for:
-        - Embedded systems (Raspberry Pi, Arduino with accelerators)
-        - Battery-powered devices
-        - Ultra-low latency requirements (<10ms)
-        - Quick training experiments
-
-    Performance (approximate):
-        - CPU inference: ~6ms (single core)
-        - Parameters: ~0.9M backbone
-        - MAdds: 56M
-
-    Args:
-        in_shape: (H, W) image dimensions
-        out_size: Number of regression targets
-        pretrained: Use ImageNet pretrained weights (default: True)
-        dropout_rate: Dropout rate in head (default: 0.2)
-        freeze_backbone: Freeze backbone for fine-tuning (default: False)
-        regression_hidden: Hidden units in regression head (default: 256)
-
-    Example:
-        >>> model = MobileNetV3Small(in_shape=(224, 224), out_size=3)
-        >>> x = torch.randn(1, 1, 224, 224)
-        >>> out = model(x)  # (1, 3)
-    """
-
-    def __init__(self, in_shape: tuple[int, int], out_size: int, **kwargs):
-        super().__init__(
-            in_shape=in_shape,
-            out_size=out_size,
-            model_fn=mobilenet_v3_small,
-            weights_class=MobileNet_V3_Small_Weights,
-            **kwargs,
-        )
-
-    def __repr__(self) -> str:
-        pt = "pretrained" if self.pretrained else "scratch"
-        return f"MobileNetV3_Small({pt}, in={self.in_shape}, out={self.out_size})"
-
-
-@register_model("mobilenet_v3_large")
-class MobileNetV3Large(MobileNetV3Base):
-    """
-    MobileNetV3-Large: Balanced efficiency and accuracy.
-
-    ~3.0M backbone parameters. Best trade-off for mobile/portable deployment.
-    Achieves ~75% ImageNet accuracy with efficient inference.
-
-    Recommended for:
-        - Mobile deployment (smartphones, tablets)
-        - Portable inspection devices
-        - Real-time processing with moderate accuracy needs
-        - Default choice for edge deployment
-
-    Performance (approximate):
-        - CPU inference: ~20ms (single core)
-        - Parameters: ~3.0M backbone
-        - MAdds: 219M
-
-    Args:
-        in_shape: (H, W) image dimensions
-        out_size: Number of regression targets
-        pretrained: Use ImageNet pretrained weights (default: True)
-        dropout_rate: Dropout rate in head (default: 0.2)
-        freeze_backbone: Freeze backbone for fine-tuning (default: False)
-        regression_hidden: Hidden units in regression head (default: 256)
-
-    Example:
-        >>> model = MobileNetV3Large(in_shape=(224, 224), out_size=3)
-        >>> x = torch.randn(1, 1, 224, 224)
-        >>> out = model(x)  # (1, 3)
-    """
-
-    def __init__(self, in_shape: tuple[int, int], out_size: int, **kwargs):
-        super().__init__(
-            in_shape=in_shape,
-            out_size=out_size,
-            model_fn=mobilenet_v3_large,
-            weights_class=MobileNet_V3_Large_Weights,
-            **kwargs,
-        )
-
-    def __repr__(self) -> str:
-        pt = "pretrained" if self.pretrained else "scratch"
-        return f"MobileNetV3_Large({pt}, in={self.in_shape}, out={self.out_size})"
+"""
+MobileNetV3: Efficient Networks for Edge Deployment
+====================================================
+
+Lightweight architecture optimized for mobile and embedded devices.
+MobileNetV3 combines neural architecture search (NAS) with hardware-aware
+optimization to achieve excellent accuracy with minimal computational cost.
+
+**Key Features**:
+    - Inverted residuals with depthwise separable convolutions
+    - Squeeze-and-Excitation (SE) attention for channel weighting
+    - h-swish activation: efficient approximation of swish
+    - Designed for real-time inference on CPUs and edge devices
+
+**Variants**:
+    - mobilenet_v3_small: Ultra-lightweight (~0.9M backbone params) - Edge/embedded
+    - mobilenet_v3_large: Balanced (~3.0M backbone params) - Mobile deployment
+
+**Use Cases**:
+    - Real-time structural health monitoring on embedded systems
+    - Field inspection with portable devices
+    - When model size and inference speed are critical
+
+**Note**: MobileNetV3 is 2D-only. For 1D data, use TCN. For 3D data, use ResNet3D.
+
+References:
+    Howard, A., et al. (2019). Searching for MobileNetV3.
+    ICCV 2019. https://arxiv.org/abs/1905.02244
+
+Author: Ductho Le (ductho.le@outlook.com)
+"""
+
+from typing import Any
+
+import torch
+import torch.nn as nn
+
+
+try:
+    from torchvision.models import (
+        MobileNet_V3_Large_Weights,
+        MobileNet_V3_Small_Weights,
+        mobilenet_v3_large,
+        mobilenet_v3_small,
+    )
+
+    MOBILENETV3_AVAILABLE = True
+except ImportError:
+    MOBILENETV3_AVAILABLE = False
+
+from wavedl.models.base import BaseModel
+from wavedl.models.registry import register_model
+
+
+class MobileNetV3Base(BaseModel):
+    """
+    Base MobileNetV3 class for regression tasks.
+
+    Wraps torchvision MobileNetV3 with:
+    - Optional pretrained weights (ImageNet-1K)
+    - Automatic input channel adaptation (grayscale → 3ch)
+    - Lightweight regression head (maintains efficiency)
+
+    MobileNetV3 is ideal for:
+    - Edge deployment (Raspberry Pi, Jetson, mobile)
+    - Real-time inference requirements
+    - Memory-constrained environments
+    - Quick prototyping and experimentation
+
+    Note: This is 2D-only. Input shape must be (H, W).
+    """
+
+    def __init__(
+        self,
+        in_shape: tuple[int, int],
+        out_size: int,
+        model_fn,
+        weights_class,
+        pretrained: bool = True,
+        dropout_rate: float = 0.2,
+        freeze_backbone: bool = False,
+        regression_hidden: int = 256,
+        **kwargs,
+    ):
+        """
+        Initialize MobileNetV3 for regression.
+
+        Args:
+            in_shape: (H, W) input image dimensions
+            out_size: Number of regression output targets
+            model_fn: torchvision model constructor
+            weights_class: Pretrained weights enum class
+            pretrained: Use ImageNet pretrained weights (default: True)
+            dropout_rate: Dropout rate in regression head (default: 0.2)
+            freeze_backbone: Freeze backbone for fine-tuning (default: False)
+            regression_hidden: Hidden units in regression head (default: 256)
+        """
+        super().__init__(in_shape, out_size)
+
+        if not MOBILENETV3_AVAILABLE:
+            raise ImportError(
+                "torchvision is required for MobileNetV3. "
+                "Install with: pip install torchvision"
+            )
+
+        if len(in_shape) != 2:
+            raise ValueError(
+                f"MobileNetV3 requires 2D input (H, W), got {len(in_shape)}D. "
+                "For 1D data, use TCN. For 3D data, use ResNet3D."
+            )
+
+        self.pretrained = pretrained
+        self.dropout_rate = dropout_rate
+        self.freeze_backbone = freeze_backbone
+        self.regression_hidden = regression_hidden
+
+        # Load pretrained backbone
+        weights = weights_class.IMAGENET1K_V1 if pretrained else None
+        self.backbone = model_fn(weights=weights)
+
+        # MobileNetV3 classifier structure:
+        # classifier[0]: Linear (features → 1280 for Large, 1024 for Small)
+        # classifier[1]: Hardswish
+        # classifier[2]: Dropout
+        # classifier[3]: Linear (1280/1024 → num_classes)
+
+        # Get the input features to the final classifier
+        in_features = self.backbone.classifier[0].in_features
+
+        # Replace classifier with lightweight regression head
+        # Keep it efficient to maintain MobileNet's speed advantage
+        self.backbone.classifier = nn.Sequential(
+            nn.Linear(in_features, regression_hidden),
+            nn.Hardswish(inplace=True),  # Match MobileNetV3's activation
+            nn.Dropout(dropout_rate),
+            nn.Linear(regression_hidden, out_size),
+        )
+
+        # Adapt first conv for single-channel input (3× memory savings vs expand)
+        self._adapt_input_channels()
+
+        # Optionally freeze backbone for fine-tuning (after adaptation so new conv is frozen too)
+        if freeze_backbone:
+            self._freeze_backbone()
+
+    def _adapt_input_channels(self):
+        """Modify first conv to accept single-channel input.
+
+        Instead of expanding 1→3 channels in forward (which triples memory),
+        we replace the first conv layer with a 1-channel version and initialize
+        weights as the mean of the pretrained RGB filters.
+        """
+        old_conv = self.backbone.features[0][0]
+        new_conv = nn.Conv2d(
+            1,  # Single channel input
+            old_conv.out_channels,
+            kernel_size=old_conv.kernel_size,
+            stride=old_conv.stride,
+            padding=old_conv.padding,
+            dilation=old_conv.dilation,
+            groups=old_conv.groups,
+            padding_mode=old_conv.padding_mode,
+            bias=old_conv.bias is not None,
+        )
+        if self.pretrained:
+            with torch.no_grad():
+                new_conv.weight.copy_(old_conv.weight.mean(dim=1, keepdim=True))
+        self.backbone.features[0][0] = new_conv
+
+    def _freeze_backbone(self):
+        """Freeze all backbone parameters except the classifier."""
+        for name, param in self.backbone.named_parameters():
+            if "classifier" not in name:
+                param.requires_grad = False
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        """
+        Forward pass.
+
+        Args:
+            x: Input tensor of shape (B, 1, H, W)
+
+        Returns:
+            Output tensor of shape (B, out_size)
+        """
+        return self.backbone(x)
+
+    @classmethod
+    def get_default_config(cls) -> dict[str, Any]:
+        """Return default configuration for MobileNetV3."""
+        return {
+            "pretrained": True,
+            "dropout_rate": 0.2,
+            "freeze_backbone": False,
+            "regression_hidden": 256,
+        }
+
+
+# =============================================================================
+# REGISTERED MODEL VARIANTS
+# =============================================================================
+
+
+@register_model("mobilenet_v3_small")
+class MobileNetV3Small(MobileNetV3Base):
+    """
+    MobileNetV3-Small: Ultra-lightweight for edge deployment.
+
+    ~0.9M backbone parameters. Designed for the most constrained environments.
+    Achieves ~67% ImageNet accuracy with minimal compute.
+
+    Recommended for:
+        - Embedded systems (Raspberry Pi, Arduino with accelerators)
+        - Battery-powered devices
+        - Ultra-low latency requirements (<10ms)
+        - Quick training experiments
+
+    Performance (approximate):
+        - CPU inference: ~6ms (single core)
+        - Parameters: ~0.9M backbone
+        - MAdds: 56M
+
+    Args:
+        in_shape: (H, W) image dimensions
+        out_size: Number of regression targets
+        pretrained: Use ImageNet pretrained weights (default: True)
+        dropout_rate: Dropout rate in head (default: 0.2)
+        freeze_backbone: Freeze backbone for fine-tuning (default: False)
+        regression_hidden: Hidden units in regression head (default: 256)
+
+    Example:
+        >>> model = MobileNetV3Small(in_shape=(224, 224), out_size=3)
+        >>> x = torch.randn(1, 1, 224, 224)
+        >>> out = model(x)  # (1, 3)
+    """
+
+    def __init__(self, in_shape: tuple[int, int], out_size: int, **kwargs):
+        super().__init__(
+            in_shape=in_shape,
+            out_size=out_size,
+            model_fn=mobilenet_v3_small,
+            weights_class=MobileNet_V3_Small_Weights,
+            **kwargs,
+        )
+
+    def __repr__(self) -> str:
+        pt = "pretrained" if self.pretrained else "scratch"
+        return f"MobileNetV3_Small({pt}, in={self.in_shape}, out={self.out_size})"
+
+
+@register_model("mobilenet_v3_large")
+class MobileNetV3Large(MobileNetV3Base):
+    """
+    MobileNetV3-Large: Balanced efficiency and accuracy.
+
+    ~3.0M backbone parameters. Best trade-off for mobile/portable deployment.
+    Achieves ~75% ImageNet accuracy with efficient inference.
+
+    Recommended for:
+        - Mobile deployment (smartphones, tablets)
+        - Portable inspection devices
+        - Real-time processing with moderate accuracy needs
+        - Default choice for edge deployment
+
+    Performance (approximate):
+        - CPU inference: ~20ms (single core)
+        - Parameters: ~3.0M backbone
+        - MAdds: 219M
+
+    Args:
+        in_shape: (H, W) image dimensions
+        out_size: Number of regression targets
+        pretrained: Use ImageNet pretrained weights (default: True)
+        dropout_rate: Dropout rate in head (default: 0.2)
+        freeze_backbone: Freeze backbone for fine-tuning (default: False)
+        regression_hidden: Hidden units in regression head (default: 256)
+
+    Example:
+        >>> model = MobileNetV3Large(in_shape=(224, 224), out_size=3)
+        >>> x = torch.randn(1, 1, 224, 224)
+        >>> out = model(x)  # (1, 3)
+    """
+
+    def __init__(self, in_shape: tuple[int, int], out_size: int, **kwargs):
+        super().__init__(
+            in_shape=in_shape,
+            out_size=out_size,
+            model_fn=mobilenet_v3_large,
+            weights_class=MobileNet_V3_Large_Weights,
+            **kwargs,
+        )
+
+    def __repr__(self) -> str:
+        pt = "pretrained" if self.pretrained else "scratch"
+        return f"MobileNetV3_Large({pt}, in={self.in_shape}, out={self.out_size})"