update input_image_shape -> image_shape (keras-team#1785)

divyashreepathihalli · mattdangerw · commit 7625cb1c1f10 · 2024-09-09T21:45:12.000-07:00
* update input_image_shape -&gt; image_shape

* update docstring example

* code reformat

* update tests
diff --git a/keras_nlp/src/models/csp_darknet/csp_darknet_backbone.py b/keras_nlp/src/models/csp_darknet/csp_darknet_backbone.py
@@ -38,7 +38,7 @@ class CSPDarkNetBackbone(Backbone):
             Use `"depthwise_block"` for depthwise conv block
             `"basic_block"` for basic conv block.
             Defaults to "basic_block".
-        input_image_shape: tuple. The input shape without the batch size.
+        image_shape: tuple. The input shape without the batch size.
             Defaults to `(None, None, 3)`.
 
     Examples:
@@ -67,7 +67,7 @@ def __init__(
         stackwise_depth,
         include_rescaling,
         block_type="basic_block",
-        input_image_shape=(224, 224, 3),
+        image_shape=(224, 224, 3),
         **kwargs,
     ):
         # === Functional Model ===
@@ -78,7 +78,7 @@ def __init__(
         )
         base_channels = stackwise_num_filters[0] // 2
 
-        image_input = layers.Input(shape=input_image_shape)
+        image_input = layers.Input(shape=image_shape)
         x = image_input
         if include_rescaling:
             x = layers.Rescaling(scale=1 / 255.0)(x)
@@ -119,7 +119,7 @@ def __init__(
         self.stackwise_depth = stackwise_depth
         self.include_rescaling = include_rescaling
         self.block_type = block_type
-        self.input_image_shape = input_image_shape
+        self.image_shape = image_shape
 
     def get_config(self):
         config = super().get_config()
@@ -129,7 +129,7 @@ def get_config(self):
                 "stackwise_depth": self.stackwise_depth,
                 "include_rescaling": self.include_rescaling,
                 "block_type": self.block_type,
-                "input_image_shape": self.input_image_shape,
+                "image_shape": self.image_shape,
             }
         )
         return config
diff --git a/keras_nlp/src/models/csp_darknet/csp_darknet_backbone_test.py b/keras_nlp/src/models/csp_darknet/csp_darknet_backbone_test.py
@@ -28,7 +28,7 @@ def setUp(self):
             "stackwise_depth": [1, 3, 3, 1],
             "include_rescaling": False,
             "block_type": "basic_block",
-            "input_image_shape": (224, 224, 3),
+            "image_shape": (224, 224, 3),
         }
         self.input_data = np.ones((2, 224, 224, 3), dtype="float32")
 
diff --git a/keras_nlp/src/models/csp_darknet/csp_darknet_image_classifier.py b/keras_nlp/src/models/csp_darknet/csp_darknet_image_classifier.py
@@ -78,7 +78,7 @@ class CSPDarkNetImageClassifier(ImageClassifier):
         stackwise_depth=[3, 9, 9, 3],
         include_rescaling=False,
         block_type="basic_block",
-        input_image_shape = (224, 224, 3),
+        image_shape = (224, 224, 3),
     )
     classifier = keras_nlp.models.CSPDarkNetImageClassifier(
         backbone=backbone,
diff --git a/keras_nlp/src/models/csp_darknet/csp_darknet_image_classifier_test.py b/keras_nlp/src/models/csp_darknet/csp_darknet_image_classifier_test.py
@@ -33,7 +33,7 @@ def setUp(self):
             stackwise_depth=[1, 3, 3, 1],
             include_rescaling=False,
             block_type="basic_block",
-            input_image_shape=(16, 16, 3),
+            image_shape=(16, 16, 3),
         )
         self.init_kwargs = {
             "backbone": self.backbone,
diff --git a/keras_nlp/src/models/densenet/densenet_backbone.py b/keras_nlp/src/models/densenet/densenet_backbone.py
@@ -35,7 +35,7 @@ class DenseNetBackbone(Backbone):
         include_rescaling: bool, whether to rescale the inputs. If set
             to `True`, inputs will be passed through a `Rescaling(1/255.0)`
             layer. Defaults to `True`.
-        input_image_shape: optional shape tuple, defaults to (224, 224, 3).
+        image_shape: optional shape tuple, defaults to (224, 224, 3).
         compression_ratio: float, compression rate at transition layers,
             defaults to 0.5.
         growth_rate: int, number of filters added by each dense block,
@@ -62,13 +62,13 @@ def __init__(
         self,
         stackwise_num_repeats,
         include_rescaling=True,
-        input_image_shape=(224, 224, 3),
+        image_shape=(224, 224, 3),
         compression_ratio=0.5,
         growth_rate=32,
         **kwargs,
     ):
         # === Functional Model ===
-        image_input = keras.layers.Input(shape=input_image_shape)
+        image_input = keras.layers.Input(shape=image_shape)
 
         x = image_input
         if include_rescaling:
@@ -116,7 +116,7 @@ def __init__(
         self.include_rescaling = include_rescaling
         self.compression_ratio = compression_ratio
         self.growth_rate = growth_rate
-        self.input_image_shape = input_image_shape
+        self.image_shape = image_shape
 
     def get_config(self):
         config = super().get_config()
@@ -126,7 +126,7 @@ def get_config(self):
                 "include_rescaling": self.include_rescaling,
                 "compression_ratio": self.compression_ratio,
                 "growth_rate": self.growth_rate,
-                "input_image_shape": self.input_image_shape,
+                "image_shape": self.image_shape,
             }
         )
         return config
diff --git a/keras_nlp/src/models/densenet/densenet_backbone_test.py b/keras_nlp/src/models/densenet/densenet_backbone_test.py
@@ -26,7 +26,7 @@ def setUp(self):
             "include_rescaling": True,
             "compression_ratio": 0.5,
             "growth_rate": 32,
-            "input_image_shape": (224, 224, 3),
+            "image_shape": (224, 224, 3),
         }
         self.input_data = np.ones((2, 224, 224, 3), dtype="float32")
 
diff --git a/keras_nlp/src/models/densenet/densenet_image_classifier.py b/keras_nlp/src/models/densenet/densenet_image_classifier.py
@@ -76,7 +76,7 @@ class DenseNetImageClassifier(ImageClassifier):
         stackwise_depth=[3, 9, 9, 3],
         include_rescaling=False,
         block_type="basic_block",
-        input_image_shape = (224, 224, 3),
+        image_shape = (224, 224, 3),
     )
     classifier = keras_nlp.models.DenseNetImageClassifier(
         backbone=backbone,
diff --git a/keras_nlp/src/models/densenet/densenet_image_classifier_test.py b/keras_nlp/src/models/densenet/densenet_image_classifier_test.py
@@ -31,7 +31,7 @@ def setUp(self):
             include_rescaling=True,
             compression_ratio=0.5,
             growth_rate=32,
-            input_image_shape=(224, 224, 3),
+            image_shape=(224, 224, 3),
         )
         self.init_kwargs = {
             "backbone": self.backbone,
diff --git a/keras_nlp/src/models/resnet/resnet_backbone.py b/keras_nlp/src/models/resnet/resnet_backbone.py
@@ -54,7 +54,7 @@ class ResNetBackbone(FeaturePyramidBackbone):
         include_rescaling: boolean. If `True`, rescale the input using
             `Rescaling` and `Normalization` layers. If `False`, do nothing.
             Defaults to `True`.
-        input_image_shape: tuple. The input shape without the batch size.
+        image_shape: tuple. The input shape without the batch size.
             Defaults to `(None, None, 3)`.
         pooling: `None` or str. Pooling mode for feature extraction. Defaults
             to `"avg"`.
@@ -107,7 +107,7 @@ def __init__(
         block_type,
         use_pre_activation=False,
         include_rescaling=True,
-        input_image_shape=(None, None, 3),
+        image_shape=(None, None, 3),
         pooling="avg",
         data_format=None,
         dtype=None,
@@ -139,7 +139,7 @@ def __init__(
         num_stacks = len(stackwise_num_filters)
 
         # === Functional Model ===
-        image_input = layers.Input(shape=input_image_shape)
+        image_input = layers.Input(shape=image_shape)
         if include_rescaling:
             x = layers.Rescaling(scale=1 / 255.0, dtype=dtype)(image_input)
             x = layers.Normalization(
@@ -254,7 +254,7 @@ def __init__(
         self.block_type = block_type
         self.use_pre_activation = use_pre_activation
         self.include_rescaling = include_rescaling
-        self.input_image_shape = input_image_shape
+        self.image_shape = image_shape
         self.pooling = pooling
         self.pyramid_outputs = pyramid_outputs
 
@@ -268,7 +268,7 @@ def get_config(self):
                 "block_type": self.block_type,
                 "use_pre_activation": self.use_pre_activation,
                 "include_rescaling": self.include_rescaling,
-                "input_image_shape": self.input_image_shape,
+                "image_shape": self.image_shape,
                 "pooling": self.pooling,
             }
         )
diff --git a/keras_nlp/src/models/resnet/resnet_backbone_test.py b/keras_nlp/src/models/resnet/resnet_backbone_test.py
@@ -27,7 +27,7 @@ def setUp(self):
             "stackwise_num_filters": [64, 64, 64],
             "stackwise_num_blocks": [2, 2, 2],
             "stackwise_num_strides": [1, 2, 2],
-            "input_image_shape": (None, None, 3),
+            "image_shape": (None, None, 3),
             "pooling": "avg",
         }
         self.input_size = 64
@@ -84,7 +84,7 @@ def test_saved_model(self, use_pre_activation, block_type):
             {
                 "block_type": block_type,
                 "use_pre_activation": use_pre_activation,
-                "input_image_shape": (None, None, 3),
+                "image_shape": (None, None, 3),
             }
         )
         self.run_model_saving_test(
diff --git a/keras_nlp/src/models/resnet/resnet_image_classifier_test.py b/keras_nlp/src/models/resnet/resnet_image_classifier_test.py
@@ -31,7 +31,7 @@ def setUp(self):
             stackwise_num_strides=[1, 2, 2],
             block_type="basic_block",
             use_pre_activation=True,
-            input_image_shape=(16, 16, 3),
+            image_shape=(16, 16, 3),
             include_rescaling=False,
             pooling="avg",
         )
diff --git a/keras_nlp/src/models/vgg/vgg_backbone.py b/keras_nlp/src/models/vgg/vgg_backbone.py
@@ -20,8 +20,7 @@
 
 @keras_nlp_export("keras_nlp.models.VGGBackbone")
 class VGGBackbone(Backbone):
-    """
-    This class represents Keras Backbone of VGG model.
+    """This class represents Keras Backbone of VGG model.
 
     This class implements a VGG backbone as described in [Very Deep
     Convolutional Networks for Large-Scale Image Recognition](
@@ -36,7 +35,7 @@ class VGGBackbone(Backbone):
             64, 128, 256, 512, 512].
       include_rescaling: bool, whether to rescale the inputs. If set to
         True, inputs will be passed through a `Rescaling(1/255.0)` layer.
-      input_shape: tuple, optional shape tuple, defaults to (224, 224, 3).
+      image_shape: tuple, optional shape tuple, defaults to (224, 224, 3).
       pooling: bool, Optional pooling mode for feature extraction
         when `include_top` is `False`.
         - `None` means that the output of the model will be
@@ -61,7 +60,7 @@ class VGGBackbone(Backbone):
     model = keras_nlp.models.VGGBackbone(
         stackwise_num_repeats = [2, 2, 3, 3, 3],
         stackwise_num_filters = [64, 128, 256, 512, 512],
-        input_shape = (224, 224, 3),
+        image_shape = (224, 224, 3),
         include_rescaling = False,
         pooling = "avg",
     )
@@ -74,13 +73,13 @@ def __init__(
         stackwise_num_repeats,
         stackwise_num_filters,
         include_rescaling,
-        input_image_shape=(224, 224, 3),
+        image_shape=(224, 224, 3),
         pooling="avg",
         **kwargs,
     ):
 
         # === Functional Model ===
-        img_input = keras.layers.Input(shape=input_image_shape)
+        img_input = keras.layers.Input(shape=image_shape)
         x = img_input
 
         if include_rescaling:
@@ -107,15 +106,15 @@ def __init__(
         self.stackwise_num_repeats = stackwise_num_repeats
         self.stackwise_num_filters = stackwise_num_filters
         self.include_rescaling = include_rescaling
-        self.input_image_shape = input_image_shape
+        self.image_shape = image_shape
         self.pooling = pooling
 
     def get_config(self):
         return {
             "stackwise_num_repeats": self.stackwise_num_repeats,
             "stackwise_num_filters": self.stackwise_num_filters,
             "include_rescaling": self.include_rescaling,
-            "input_image_shape": self.input_image_shape,
+            "image_shape": self.image_shape,
             "pooling": self.pooling,
         }
 
diff --git a/keras_nlp/src/models/vgg/vgg_backbone_test.py b/keras_nlp/src/models/vgg/vgg_backbone_test.py
@@ -24,7 +24,7 @@ def setUp(self):
         self.init_kwargs = {
             "stackwise_num_repeats": [2, 3, 3],
             "stackwise_num_filters": [8, 64, 64],
-            "input_image_shape": (16, 16, 3),
+            "image_shape": (16, 16, 3),
             "include_rescaling": False,
             "pooling": "avg",
         }
diff --git a/keras_nlp/src/models/vgg/vgg_image_classifier.py b/keras_nlp/src/models/vgg/vgg_image_classifier.py
@@ -65,7 +65,7 @@ class VGGImageClassifier(ImageClassifier):
     backbone = keras_nlp.models.VGGBackbone(
         stackwise_num_repeats = [2, 2, 3, 3, 3],
         stackwise_num_filters = [64, 128, 256, 512, 512],
-        input_shape = (224, 224, 3),
+        image_shape = (224, 224, 3),
         include_rescaling = False,
         pooling = "avg",
     )
diff --git a/keras_nlp/src/models/vgg/vgg_image_classifier_test.py b/keras_nlp/src/models/vgg/vgg_image_classifier_test.py
@@ -27,7 +27,7 @@ def setUp(self):
         self.backbone = VGGBackbone(
             stackwise_num_repeats=[2, 4, 4],
             stackwise_num_filters=[2, 16, 16],
-            input_image_shape=(4, 4, 3),
+            image_shape=(4, 4, 3),
             include_rescaling=False,
             pooling="max",
         )
diff --git a/keras_nlp/src/tests/test_case.py b/keras_nlp/src/tests/test_case.py
@@ -501,10 +501,10 @@ def run_vision_backbone_test(
                 input_data = ops.transpose(input_data, axes=(2, 0, 1))
             elif len(input_data_shape) == 4:
                 input_data = ops.transpose(input_data, axes=(0, 3, 1, 2))
-            if "input_image_shape" in init_kwargs:
+            if "image_shape" in init_kwargs:
                 init_kwargs = init_kwargs.copy()
-                init_kwargs["input_image_shape"] = tuple(
-                    reversed(init_kwargs["input_image_shape"])
+                init_kwargs["image_shape"] = tuple(
+                    reversed(init_kwargs["image_shape"])
                 )
             self.run_backbone_test(
                 cls=cls,

Original file line number	Diff line number	Diff line change
`@@ -28,7 +28,7 @@ def setUp(self):`
`28`	`28`	`"stackwise_depth": [1, 3, 3, 1],`
`29`	`29`	`"include_rescaling": False,`
`30`	`30`	`"block_type": "basic_block",`
`31`		`- "input_image_shape": (224, 224, 3),`
	`31`	`+ "image_shape": (224, 224, 3),`
`32`	`32`	`}`
`33`	`33`	`self.input_data = np.ones((2, 224, 224, 3), dtype="float32")`
`34`	`34`
Original file line number	Diff line number	Diff line change
`@@ -78,7 +78,7 @@ class CSPDarkNetImageClassifier(ImageClassifier):`
`78`	`78`	`stackwise_depth=[3, 9, 9, 3],`
`79`	`79`	`include_rescaling=False,`
`80`	`80`	`block_type="basic_block",`
`81`		`- input_image_shape = (224, 224, 3),`
	`81`	`+ image_shape = (224, 224, 3),`
`82`	`82`	`)`
`83`	`83`	`classifier = keras_nlp.models.CSPDarkNetImageClassifier(`
`84`	`84`	`backbone=backbone,`
Original file line number	Diff line number	Diff line change
`@@ -33,7 +33,7 @@ def setUp(self):`
`33`	`33`	`stackwise_depth=[1, 3, 3, 1],`
`34`	`34`	`include_rescaling=False,`
`35`	`35`	`block_type="basic_block",`
`36`		`- input_image_shape=(16, 16, 3),`
	`36`	`+ image_shape=(16, 16, 3),`
`37`	`37`	`)`
`38`	`38`	`self.init_kwargs = {`
`39`	`39`	`"backbone": self.backbone,`
Original file line number	Diff line number	Diff line change
`@@ -26,7 +26,7 @@ def setUp(self):`
`26`	`26`	`"include_rescaling": True,`
`27`	`27`	`"compression_ratio": 0.5,`
`28`	`28`	`"growth_rate": 32,`
`29`		`- "input_image_shape": (224, 224, 3),`
	`29`	`+ "image_shape": (224, 224, 3),`
`30`	`30`	`}`
`31`	`31`	`self.input_data = np.ones((2, 224, 224, 3), dtype="float32")`
`32`	`32`
Original file line number	Diff line number	Diff line change
`@@ -76,7 +76,7 @@ class DenseNetImageClassifier(ImageClassifier):`
`76`	`76`	`stackwise_depth=[3, 9, 9, 3],`
`77`	`77`	`include_rescaling=False,`
`78`	`78`	`block_type="basic_block",`
`79`		`- input_image_shape = (224, 224, 3),`
	`79`	`+ image_shape = (224, 224, 3),`
`80`	`80`	`)`
`81`	`81`	`classifier = keras_nlp.models.DenseNetImageClassifier(`
`82`	`82`	`backbone=backbone,`
Original file line number	Diff line number	Diff line change
`@@ -31,7 +31,7 @@ def setUp(self):`
`31`	`31`	`include_rescaling=True,`
`32`	`32`	`compression_ratio=0.5,`
`33`	`33`	`growth_rate=32,`
`34`		`- input_image_shape=(224, 224, 3),`
	`34`	`+ image_shape=(224, 224, 3),`
`35`	`35`	`)`
`36`	`36`	`self.init_kwargs = {`
`37`	`37`	`"backbone": self.backbone,`
Original file line number	Diff line number	Diff line change
`@@ -27,7 +27,7 @@ def setUp(self):`
`27`	`27`	`"stackwise_num_filters": [64, 64, 64],`
`28`	`28`	`"stackwise_num_blocks": [2, 2, 2],`
`29`	`29`	`"stackwise_num_strides": [1, 2, 2],`
`30`		`- "input_image_shape": (None, None, 3),`
	`30`	`+ "image_shape": (None, None, 3),`
`31`	`31`	`"pooling": "avg",`
`32`	`32`	`}`
`33`	`33`	`self.input_size = 64`
`@@ -84,7 +84,7 @@ def test_saved_model(self, use_pre_activation, block_type):`
`84`	`84`	`{`
`85`	`85`	`"block_type": block_type,`
`86`	`86`	`"use_pre_activation": use_pre_activation,`
`87`		`- "input_image_shape": (None, None, 3),`
	`87`	`+ "image_shape": (None, None, 3),`
`88`	`88`	`}`
`89`	`89`	`)`
`90`	`90`	`self.run_model_saving_test(`