segmodels-keras 0.1.0.dev5__tar.gz → 0.1.0.dev7__tar.gz

{segmodels_keras-0.1.0.dev5 → segmodels_keras-0.1.0.dev7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: segmodels_keras
-Version: 0.1.0.dev5
+Version: 0.1.0.dev7
 Summary: Image segmentation models with pre-trained backbones with Keras.
 Home-page: https://github.com/orthoseg/segmodels_keras
 Author: Pieter Roggemans

{segmodels_keras-0.1.0.dev5 → segmodels_keras-0.1.0.dev7}/segmodels_keras/__init__.py

@@ -59,17 +59,17 @@ def set_framework(name):
     name = name.lower()
 
     if name == _KERAS_FRAMEWORK_NAME:
-        import keras
+        import keras  # noqa: PLC0415
     elif name == _TF_KERAS_FRAMEWORK_NAME:
-        from tensorflow import keras
+        from tensorflow import keras  # noqa: PLC0415
     else:
         raise ValueError(
             f"Not correct module name `{name}`, use `{_KERAS_FRAMEWORK_NAME}` or "
             f"`{_TF_KERAS_FRAMEWORK_NAME}`"
         )
 
-    global _KERAS_BACKEND, _KERAS_LAYERS, _KERAS_MODELS
-    global _KERAS_UTILS, _KERAS_LOSSES, _KERAS_FRAMEWORK
+    global _KERAS_BACKEND, _KERAS_LAYERS, _KERAS_MODELS  # noqa: PLW0603
+    global _KERAS_UTILS, _KERAS_LOSSES, _KERAS_FRAMEWORK  # noqa: PLW0603
 
     _KERAS_FRAMEWORK = name
     _KERAS_BACKEND = keras.backend

{segmodels_keras-0.1.0.dev5 → segmodels_keras-0.1.0.dev7}/segmodels_keras/_version.py

@@ -1,2 +1,2 @@
-__versioninfo__ = (0, 1, 0, "dev5")
+__versioninfo__ = (0, 1, 0, "dev7")
 __version__ = ".".join(map(str, __versioninfo__))

segmodels_keras-0.1.0.dev7/segmodels_keras/backbones/_common_blocks.py

@@ -0,0 +1,131 @@
+"""Common blocks for backbone models.
+
+Source: https://github.com/qubvel/classification_models
+"""
+
+from keras import backend, layers
+
+
+def slice_tensor(x, start, stop, axis):
+    if axis == 3:
+        return x[:, :, :, start:stop]
+    elif axis == 1:
+        return x[:, start:stop, :, :]
+    else:
+        raise ValueError(f"Slice axis should be in (1, 3), got {axis}.")
+
+
+def GroupConv2D(
+    filters,
+    kernel_size,
+    strides=(1, 1),
+    groups=32,
+    kernel_initializer="he_uniform",
+    use_bias=True,
+    activation="linear",
+    padding="valid",
+    **kwargs,  # noqa: ARG001
+):
+    """Grouped Convolution Layer implemented as a Slice, Conv2D and Concatenate layers.
+
+    Split filters to groups, apply Conv2D and concatenate back.
+
+    Args:
+        filters: Integer, the dimensionality of the output space
+            (i.e. the number of output filters in the convolution).
+        kernel_size: An integer or tuple/list of a single integer,
+            specifying the length of the 1D convolution window.
+        strides: An integer or tuple/list of a single integer, specifying the stride
+            length of the convolution.
+        groups: Integer, number of groups to split input filters to.
+        kernel_initializer: Regularizer function applied to the kernel weights matrix.
+        use_bias: Boolean, whether the layer uses a bias vector.
+        activation: Activation function to use (see activations).
+            If you don't specify anything, no activation is applied
+            (ie. "linear" activation: a(x) = x).
+        padding: one of "valid" or "same" (case-insensitive).
+
+    Input shape:
+        4D tensor with shape: (batch, rows, cols, channels) if data_format is
+        "channels_last".
+
+    Output shape:
+        4D tensor with shape: (batch, new_rows, new_cols, filters) if data_format is
+        "channels_last". rows and cols values might have changed due to padding.
+
+    """
+    slice_axis = 3 if backend.image_data_format() == "channels_last" else 1
+
+    def layer(input_tensor):
+        inp_ch = int(
+            backend.int_shape(input_tensor)[-1] // groups
+        )  # input grouped channels
+        out_ch = int(filters // groups)  # output grouped channels
+
+        blocks = []
+        for c in range(groups):
+            slice_arguments = {
+                "start": c * inp_ch,
+                "stop": (c + 1) * inp_ch,
+                "axis": slice_axis,
+            }
+            x = layers.Lambda(slice_tensor, arguments=slice_arguments)(input_tensor)
+            x = layers.Conv2D(
+                out_ch,
+                kernel_size,
+                strides=strides,
+                kernel_initializer=kernel_initializer,
+                use_bias=use_bias,
+                activation=activation,
+                padding=padding,
+            )(x)
+            blocks.append(x)
+
+        x = layers.Concatenate(axis=slice_axis)(blocks)
+        return x
+
+    return layer
+
+
+def expand_dims(x, channels_axis):
+    if channels_axis == 3:
+        return x[:, None, None, :]
+    elif channels_axis == 1:
+        return x[:, :, None, None]
+    else:
+        raise ValueError(f"Slice axis should be in (1, 3), got {channels_axis}.")
+
+
+def ChannelSE(reduction=16, **kwargs):  # noqa: ARG001
+    """
+    Squeeze and Excitation block, reimplementation inspired by
+        https://github.com/Cadene/pretrained-models.pytorch/blob/master/pretrainedmodels/models/senet.py
+
+    Args:
+        reduction: channels squeeze factor
+
+    """
+    channels_axis = 3 if backend.image_data_format() == "channels_last" else 1
+
+    def layer(input_tensor):
+        # get number of channels/filters
+        channels = backend.int_shape(input_tensor)[channels_axis]
+
+        x = input_tensor
+
+        # squeeze and excitation block in PyTorch style with
+        x = layers.GlobalAveragePooling2D()(x)
+        x = layers.Lambda(expand_dims, arguments={"channels_axis": channels_axis})(x)
+        x = layers.Conv2D(
+            channels // reduction, (1, 1), kernel_initializer="he_uniform"
+        )(x)
+        x = layers.Activation("relu")(x)
+        x = layers.Conv2D(channels, (1, 1), kernel_initializer="he_uniform")(x)
+        x = layers.Activation("sigmoid")(x)
+
+        # apply attention
+        x = layers.Multiply()([input_tensor, x])
+
+        return x
+
+    return layer

segmodels_keras-0.1.0.dev7/segmodels_keras/backbones/_weights.py

@@ -0,0 +1,363 @@
+"""Helper to download weights for backbones.
+
+Source: https://github.com/qubvel/classification_models
+"""
+
+from keras.utils import get_file
+
+__all__ = ["load_model_weights"]
+
+
+def _find_weights(model_name, dataset, include_top):
+    w = list(filter(lambda x: x["model"] == model_name, WEIGHTS_COLLECTION))
+    w = list(filter(lambda x: x["dataset"] == dataset, w))
+    w = list(filter(lambda x: x["include_top"] == include_top, w))
+    return w
+
+
+def load_model_weights(model, model_name, dataset, classes, include_top):
+
+    weights = _find_weights(model_name, dataset, include_top)
+
+    if weights:
+        weights = weights[0]
+
+        if include_top and weights["classes"] != classes:
+            raise ValueError(
+                "If using `weights` and `include_top`"
+                " as true, `classes` should be {}".format(weights["classes"])
+            )
+
+        weights_path = get_file(
+            weights["name"],
+            weights["url"],
+            cache_subdir="models",
+            md5_hash=weights["md5"],
+        )
+
+        model.load_weights(weights_path)
+
+    else:
+        raise ValueError(
+            "There is no weights for such configuration: "
+            f"model = {model.name}, dataset = {dataset}, "
+            f"classes = {classes}, include_top = {include_top}."
+        )
+
+
+WEIGHTS_COLLECTION = [
+    # ResNet18
+    {
+        "model": "resnet18",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet18_imagenet_1000.h5",
+        "name": "resnet18_imagenet_1000.h5",
+        "md5": "64da73012bb70e16c901316c201d9803",
+    },
+    {
+        "model": "resnet18",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet18_imagenet_1000_no_top.h5",
+        "name": "resnet18_imagenet_1000_no_top.h5",
+        "md5": "318e3ac0cd98d51e917526c9f62f0b50",
+    },
+    # ResNet34
+    {
+        "model": "resnet34",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet34_imagenet_1000.h5",
+        "name": "resnet34_imagenet_1000.h5",
+        "md5": "2ac8277412f65e5d047f255bcbd10383",
+    },
+    {
+        "model": "resnet34",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet34_imagenet_1000_no_top.h5",
+        "name": "resnet34_imagenet_1000_no_top.h5",
+        "md5": "8caaa0ad39d927cb8ba5385bf945d582",
+    },
+    # ResNet50
+    {
+        "model": "resnet50",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet50_imagenet_1000.h5",
+        "name": "resnet50_imagenet_1000.h5",
+        "md5": "d0feba4fc650e68ac8c19166ee1ba87f",
+    },
+    {
+        "model": "resnet50",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet50_imagenet_1000_no_top.h5",
+        "name": "resnet50_imagenet_1000_no_top.h5",
+        "md5": "db3b217156506944570ac220086f09b6",
+    },
+    {
+        "model": "resnet50",
+        "dataset": "imagenet11k-places365ch",
+        "classes": 11586,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet50_imagenet11k-places365ch_11586.h5",
+        "name": "resnet50_imagenet11k-places365ch_11586.h5",
+        "md5": "bb8963db145bc9906452b3d9c9917275",
+    },
+    {
+        "model": "resnet50",
+        "dataset": "imagenet11k-places365ch",
+        "classes": 11586,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet50_imagenet11k-places365ch_11586_no_top.h5",
+        "name": "resnet50_imagenet11k-places365ch_11586_no_top.h5",
+        "md5": "d8bf4e7ea082d9d43e37644da217324a",
+    },
+    # ResNet101
+    {
+        "model": "resnet101",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet101_imagenet_1000.h5",
+        "name": "resnet101_imagenet_1000.h5",
+        "md5": "9489ed2d5d0037538134c880167622ad",
+    },
+    {
+        "model": "resnet101",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet101_imagenet_1000_no_top.h5",
+        "name": "resnet101_imagenet_1000_no_top.h5",
+        "md5": "1016e7663980d5597a4e224d915c342d",
+    },
+    # ResNet152
+    {
+        "model": "resnet152",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet152_imagenet_1000.h5",
+        "name": "resnet152_imagenet_1000.h5",
+        "md5": "1efffbcc0708fb0d46a9d096ae14f905",
+    },
+    {
+        "model": "resnet152",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet152_imagenet_1000_no_top.h5",
+        "name": "resnet152_imagenet_1000_no_top.h5",
+        "md5": "5867b94098df4640918941115db93734",
+    },
+    {
+        "model": "resnet152",
+        "dataset": "imagenet11k",
+        "classes": 11221,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet152_imagenet11k_11221.h5",
+        "name": "resnet152_imagenet11k_11221.h5",
+        "md5": "24791790f6ef32f274430ce4a2ffee5d",
+    },
+    {
+        "model": "resnet152",
+        "dataset": "imagenet11k",
+        "classes": 11221,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet152_imagenet11k_11221_no_top.h5",
+        "name": "resnet152_imagenet11k_11221_no_top.h5",
+        "md5": "25ab66dec217cb774a27d0f3659cafb3",
+    },
+    # ResNeXt50
+    {
+        "model": "resnext50",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnext50_imagenet_1000.h5",
+        "name": "resnext50_imagenet_1000.h5",
+        "md5": "7c5c40381efb044a8dea5287ab2c83db",
+    },
+    {
+        "model": "resnext50",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnext50_imagenet_1000_no_top.h5",
+        "name": "resnext50_imagenet_1000_no_top.h5",
+        "md5": "7ade5c8aac9194af79b1724229bdaa50",
+    },
+    # ResNeXt101
+    {
+        "model": "resnext101",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnext101_imagenet_1000.h5",
+        "name": "resnext101_imagenet_1000.h5",
+        "md5": "432536e85ee811568a0851c328182735",
+    },
+    {
+        "model": "resnext101",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/resnext101_imagenet_1000_no_top.h5",
+        "name": "resnext101_imagenet_1000_no_top.h5",
+        "md5": "91fe0126320e49f6ee607a0719828c7e",
+    },
+    # SE models
+    {
+        "model": "seresnet50",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet50_imagenet_1000.h5",
+        "name": "seresnet50_imagenet_1000.h5",
+        "md5": "ff0ce1ed5accaad05d113ecef2d29149",
+    },
+    {
+        "model": "seresnet50",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet50_imagenet_1000_no_top.h5",
+        "name": "seresnet50_imagenet_1000_no_top.h5",
+        "md5": "043777781b0d5ca756474d60bf115ef1",
+    },
+    {
+        "model": "seresnet101",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet101_imagenet_1000.h5",
+        "name": "seresnet101_imagenet_1000.h5",
+        "md5": "5c31adee48c82a66a32dee3d442f5be8",
+    },
+    {
+        "model": "seresnet101",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet101_imagenet_1000_no_top.h5",
+        "name": "seresnet101_imagenet_1000_no_top.h5",
+        "md5": "1c373b0c196918713da86951d1239007",
+    },
+    {
+        "model": "seresnet152",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet152_imagenet_1000.h5",
+        "name": "seresnet152_imagenet_1000.h5",
+        "md5": "96fc14e3a939d4627b0174a0e80c7371",
+    },
+    {
+        "model": "seresnet152",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet152_imagenet_1000_no_top.h5",
+        "name": "seresnet152_imagenet_1000_no_top.h5",
+        "md5": "f58d4c1a511c7445ab9a2c2b83ee4e7b",
+    },
+    {
+        "model": "seresnext50",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnext50_imagenet_1000.h5",
+        "name": "seresnext50_imagenet_1000.h5",
+        "md5": "5310dcd58ed573aecdab99f8df1121d5",
+    },
+    {
+        "model": "seresnext50",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnext50_imagenet_1000_no_top.h5",
+        "name": "seresnext50_imagenet_1000_no_top.h5",
+        "md5": "b0f23d2e1cd406d67335fb92d85cc279",
+    },
+    {
+        "model": "seresnext101",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnext101_imagenet_1000.h5",
+        "name": "seresnext101_imagenet_1000.h5",
+        "md5": "be5b26b697a0f7f11efaa1bb6272fc84",
+    },
+    {
+        "model": "seresnext101",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnext101_imagenet_1000_no_top.h5",
+        "name": "seresnext101_imagenet_1000_no_top.h5",
+        "md5": "e48708cbe40071cc3356016c37f6c9c7",
+    },
+    {
+        "model": "senet154",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/senet154_imagenet_1000.h5",
+        "name": "senet154_imagenet_1000.h5",
+        "md5": "c8eac0e1940ea4d8a2e0b2eb0cdf4e75",
+    },
+    {
+        "model": "senet154",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/senet154_imagenet_1000_no_top.h5",
+        "name": "senet154_imagenet_1000_no_top.h5",
+        "md5": "d854ff2cd7e6a87b05a8124cd283e0f2",
+    },
+    {
+        "model": "seresnet18",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet18_imagenet_1000.h5",
+        "name": "seresnet18_imagenet_1000.h5",
+        "md5": "9a925fd96d050dbf7cc4c54aabfcf749",
+    },
+    {
+        "model": "seresnet18",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet18_imagenet_1000_no_top.h5",
+        "name": "seresnet18_imagenet_1000_no_top.h5",
+        "md5": "a46e5cd4114ac946ecdc58741e8d92ea",
+    },
+    {
+        "model": "seresnet34",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": True,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet34_imagenet_1000.h5",
+        "name": "seresnet34_imagenet_1000.h5",
+        "md5": "863976b3bd439ff0cc05c91821218a6b",
+    },
+    {
+        "model": "seresnet34",
+        "dataset": "imagenet",
+        "classes": 1000,
+        "include_top": False,
+        "url": "https://github.com/qubvel/classification_models/releases/download/0.0.1/seresnet34_imagenet_1000_no_top.h5",
+        "name": "seresnet34_imagenet_1000_no_top.h5",
+        "md5": "3348fd049f1f9ad307c070ff2b6ec4cb",
+    },
+]

{segmodels_keras-0.1.0.dev5 → segmodels_keras-0.1.0.dev7}/segmodels_keras/backbones/backbones_factory.py

@@ -2,6 +2,7 @@ import keras.applications as ka
 
 from . import inception_resnet_v2 as irv2
 from . import inception_v3 as iv3
+from . import resnet
 
 
 class BackbonesFactory:
@@ -16,39 +17,37 @@ class BackbonesFactory:
     """
 
     _models = {  # noqa: RUF012
-        # ResNets
+        # ResNets < 50 layers are NOT available via keras.applications, so use
+        # implementation from classification_models.
+        "resnet18": (
+            resnet.ResNet18,
+            resnet.preprocess_input,
+            ("stage4_unit1_relu1", "stage3_unit1_relu1", "stage2_unit1_relu1", "relu0"),
+        ),
+        "resnet34": (
+            resnet.ResNet34,
+            resnet.preprocess_input,
+            ("stage4_unit1_relu1", "stage3_unit1_relu1", "stage2_unit1_relu1", "relu0"),
+        ),
+        # ResNets > 50 layers are available via keras.applications, so use those.
+        # Skip layers (inverted) from https://github.com/yingkaisha/keras-unet-collection
         "resnet50": (
             ka.ResNet50,
             ka.resnet.preprocess_input,
-            (
-                "conv4_block6_out",
-                "conv3_block4_out",
-                "conv2_block3_out",
-                "conv1_relu",
-            ),
+            ("conv4_block6_out", "conv3_block4_out", "conv2_block3_out", "conv1_relu"),
         ),
         "resnet101": (
             ka.ResNet101,
             ka.resnet.preprocess_input,
-            (
-                "conv4_block23_out",
-                "conv3_block4_out",
-                "conv2_block3_out",
-                "conv1_relu",
-            ),
+            ("conv4_block23_out", "conv3_block4_out", "conv2_block3_out", "conv1_relu"),
         ),
         "resnet152": (
             ka.ResNet152,
             ka.resnet.preprocess_input,
-            (
-                "conv4_block36_out",
-                "conv3_block8_out",
-                "conv2_block3_out",
-                "conv1_relu",
-            ),
+            ("conv4_block36_out", "conv3_block8_out", "conv2_block3_out", "conv1_relu"),
         ),
         # ResNetV2
-        # Layers reused from: https://github.com/yingkaisha/keras-unet-collection
+        # Skip layers (inverted) from https://github.com/yingkaisha/keras-unet-collection
         "resnet50v2": (
             ka.ResNet50V2,
             ka.resnet_v2.preprocess_input,
@@ -80,6 +79,7 @@ class BackbonesFactory:
             ),
         ),
         # VGG
+        # Skip layers from segmentation_models
         "vgg16": (
             ka.vgg16.VGG16,
             ka.vgg16.preprocess_input,
@@ -103,6 +103,7 @@ class BackbonesFactory:
             ),
         ),
         # DenseNet
+        # Skip layers (inverted) from https://github.com/yingkaisha/keras-unet-collection
         "densenet121": (
             ka.densenet.DenseNet121,
             ka.densenet.preprocess_input,
@@ -122,6 +123,7 @@ class BackbonesFactory:
             ("pool4_conv", "pool3_conv", "pool2_conv", "conv1/relu"),
         ),
         # Inception
+        # Skip layers from segmentation_models
         "inceptionresnetv2": (
             irv2.InceptionResNetV2,
             irv2.preprocess_input,
@@ -138,6 +140,7 @@ class BackbonesFactory:
         ),
         "inceptionv3": (iv3.InceptionV3, iv3.preprocess_input, (228, 86, 16, 9)),
         # MobileNet
+        # Skip layers from segmentation_models
         "mobilenet": (
             ka.mobilenet.MobileNet,
             ka.mobilenet.preprocess_input,
@@ -158,6 +161,8 @@ class BackbonesFactory:
                 "block_1_expand_relu",
             ),
         ),
+        # EfficientNet
+        # Skip layers from segmentation_models
         "efficientnetb0": [
             ka.EfficientNetB0,
             ka.efficientnet.preprocess_input,
@@ -239,7 +244,7 @@ class BackbonesFactory:
             ),
         ],
         # EfficientNetV2
-        # Layers reused from: https://github.com/chinefed/segmentation_models_fork
+        # Skip layers from https://github.com/chinefed/segmentation_models_fork
         "efficientnetv2m": (
             ka.EfficientNetV2M,
             ka.efficientnet_v2.preprocess_input,

{segmodels_keras-0.1.0.dev5 → segmodels_keras-0.1.0.dev7}/segmodels_keras/backbones/inception_v3.py

@@ -8,6 +8,7 @@ and that the input preprocessing function is also different (same as Xception).
 """
 
 import os
+import warnings
 
 from keras import backend, layers, models
 from keras import utils as keras_utils