[Style]: Adjust style

paddlers/models/cd/models/__init__.py

@@ -12,12 +12,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from .bit import BIT
 from .cdnet import CDNet
-from .unet_ef import UNetEarlyFusion
+from .dsifn import DSIFN
-from .unet_siamconc import UNetSiamConc
-from .unet_siamdiff import UNetSiamDiff
 from .stanet import STANet
-from .bit import BIT
 from .snunet import SNUNet
-from .dsifn import DSIFN
 from .dsamnet import DSAMNet
+from .unet_ef import UNetEarlyFusion
+from .unet_siamconc import UNetSiamConc
+from .unet_siamdiff import UNetSiamDiff

paddlers/models/cd/models/bit.py

@@ -29,7 +29,7 @@ class BIT(nn.Layer):
 
     The original article refers to
         H. Chen, et al., "Remote Sensing Image Change Detection With Transformers"
-    (https://arxiv.org/abs/2103.00208)
+        (https://arxiv.org/abs/2103.00208).
 
     This implementation adopts pretrained encoders, as opposed to the original work where weights are randomly initialized.
 
@@ -56,6 +56,7 @@ class BIT(nn.Layer):
     Raises:
         ValueError: When an unsupported backbone type is specified, or the number of backbone stages is not 3, 4, or 5.
     """
+    
     def __init__(
         self, in_channels, num_classes,
         backbone='resnet18', n_stages=4, 

paddlers/models/cd/models/dsamnet.py

@@ -28,7 +28,7 @@ class DSAMNet(nn.Layer):
     The original article refers to
         Q. Shi, et al., "A Deeply Supervised Attention Metric-Based Network and an Open Aerial Image Dataset for Remote Sensing 
         Change Detection"
-    (https://ieeexplore.ieee.org/document/9467555)
+        (https://ieeexplore.ieee.org/document/9467555).
 
     Note that this implementation differs from the original work in two aspects:
     1. We do not use multiple dilation rates in layer 4 of the ResNet backbone.
@@ -40,6 +40,7 @@ class DSAMNet(nn.Layer):
         ca_ratio (int, optional): The channel reduction ratio for the channel attention module. Default: 8.
         sa_kernel (int, optional): The size of the convolutional kernel used in the spatial attention module. Default: 7.
     """
+    
     def __init__(self, in_channels, num_classes, ca_ratio=8, sa_kernel=7):
         super().__init__()
 

paddlers/models/cd/models/dsifn.py

@@ -28,7 +28,7 @@ class DSIFN(nn.Layer):
     The original article refers to
         C. Zhang, et al., "A deeply supervised image fusion network for change detection in high resolution bi-temporal remote 
         sensing images"
-    (https://www.sciencedirect.com/science/article/pii/S0924271620301532)
+        (https://www.sciencedirect.com/science/article/pii/S0924271620301532).
 
     Note that in this implementation, there is a flexible number of target classes.
 
@@ -37,6 +37,7 @@ class DSIFN(nn.Layer):
         use_dropout (bool, optional): A bool value that indicates whether to use dropout layers. When the model is trained 
             on a relatively small dataset, the dropout layers help prevent overfitting. Default: False.
     """
+
     def __init__(self, num_classes, use_dropout=False):
         super().__init__()
 

paddlers/models/cd/models/layers/attention.py

@@ -26,12 +26,13 @@ class ChannelAttention(nn.Layer):
 
     The original article refers to 
         Sanghyun Woo, et al., "CBAM: Convolutional Block Attention Module"
-    (https://arxiv.org/abs/1807.06521)
+        (https://arxiv.org/abs/1807.06521).
 
     Args:
         in_ch (int): The number of channels of the input features.
         ratio (int, optional): The channel reduction ratio. Default: 8.
     """
+
     def __init__(self, in_ch, ratio=8):
         super().__init__()
         self.avg_pool = nn.AdaptiveAvgPool2D(1)
@@ -52,11 +53,12 @@ class SpatialAttention(nn.Layer):
 
     The original article refers to 
         Sanghyun Woo, et al., "CBAM: Convolutional Block Attention Module"
-    (https://arxiv.org/abs/1807.06521)
+        (https://arxiv.org/abs/1807.06521).
 
     Args:
         kernel_size (int, optional): The size of the convolutional kernel. Default: 7.
     """
+
     def __init__(self, kernel_size=7):
         super().__init__()
         self.conv = BasicConv(2, 1, kernel_size, bias=False)
@@ -75,13 +77,14 @@ class CBAM(nn.Layer):
 
     The original article refers to 
         Sanghyun Woo, et al., "CBAM: Convolutional Block Attention Module"
-    (https://arxiv.org/abs/1807.06521)
+        (https://arxiv.org/abs/1807.06521).
 
     Args:
         in_ch (int): The number of channels of the input features.
         ratio (int, optional): The channel reduction ratio for the channel attention module. Default: 8.
         kernel_size (int, optional): The size of the convolutional kernel used in the spatial attention module. Default: 7.
     """
+    
     def __init__(self, in_ch, ratio=8, kernel_size=7):
         super().__init__()
         self.ca = ChannelAttention(in_ch, ratio=ratio)

paddlers/models/cd/models/param_init.py

@@ -26,6 +26,7 @@ def normal_init(param, *args, **kwargs):
     Returns:
         The initialized parameters.
     """
+    
     return nn.initializer.Normal(*args, **kwargs)(param)
 
 
@@ -42,6 +43,7 @@ def kaiming_normal_init(param, *args, **kwargs):
     Returns:
         The initialized parameters.
     """
+
     return nn.initializer.KaimingNormal(*args, **kwargs)(param)
 
 
@@ -55,6 +57,7 @@ def constant_init(param, *args, **kwargs):
     Returns:
         The initialized parameters.
     """
+
     return nn.initializer.Constant(*args, **kwargs)(param)
 
 
@@ -73,6 +76,7 @@ class KaimingInitMixin:
                 self.bn = nn.BatchNorm2D(num_classes)
                 self.init_weight()
     """
+
     def init_weight(self):
         for layer in self.sublayers():
             if isinstance(layer, nn.Conv2D):

paddlers/models/cd/models/snunet.py

@@ -28,7 +28,7 @@ class SNUNet(nn.Layer, KaimingInitMixin):
 
     The original article refers to
         S. Fang, et al., "SNUNet-CD: A Densely Connected Siamese Network for Change Detection of VHR Images"
-    (https://ieeexplore.ieee.org/document/9355573)
+        (https://ieeexplore.ieee.org/document/9355573).
 
     Note that bilinear interpolation is adopted as the upsampling method, which is different from the paper.
 
@@ -37,6 +37,7 @@ class SNUNet(nn.Layer, KaimingInitMixin):
         num_classes (int): The number of target classes.
         width (int, optional): The output channels of the first convolutional layer. Default: 32.
     """
+    
     def __init__(self, in_channels, num_classes, width=32):
         super().__init__()
 

paddlers/models/cd/models/stanet.py

@@ -28,7 +28,7 @@ class STANet(nn.Layer):
 
     The original article refers to
         H. Chen and Z. Shi, "A Spatial-Temporal Attention-Based Method and a New Dataset for Remote Sensing Image Change Detection"
-    (https://www.mdpi.com/2072-4292/12/10/1662)
+        (https://www.mdpi.com/2072-4292/12/10/1662).
 
     Note that this implementation differs from the original work in two aspects:
     1. We do not use multiple dilation rates in layer 4 of the ResNet backbone.
@@ -45,6 +45,7 @@ class STANet(nn.Layer):
     Raises:
         ValueError: When `att_type` has an illeagal value (unsupported attention type).
     """
+    
     def __init__(
         self, 
         in_channels, 

paddlers/models/cd/models/unet_ef.py

@@ -27,15 +27,15 @@ class UNetEarlyFusion(nn.Layer):
 
     The original article refers to
         Caye Daudt, R., et al. "Fully convolutional siamese networks for change detection"
-    (https://arxiv.org/abs/1810.08462)
+        (https://arxiv.org/abs/1810.08462).
 
     Args:
         in_channels (int): The number of bands of the input images.
         num_classes (int): The number of target classes.
-        use_dropout (bool, optional): A bool value that indicates whether to use dropout layers. 
+        use_dropout (bool, optional): A bool value that indicates whether to use dropout layers. When the model is trained 
-            When the model is trained on a relatively small dataset, the dropout layers help prevent 
+            on a relatively small dataset, the dropout layers help prevent overfitting. Default: False.
-            overfitting. Default: False.
     """
+
     def __init__(
         self, 
         in_channels, 

paddlers/models/cd/models/unet_siamconc.py

@@ -27,15 +27,15 @@ class UNetSiamConc(nn.Layer):
 
     The original article refers to
         Caye Daudt, R., et al. "Fully convolutional siamese networks for change detection"
-    (https://arxiv.org/abs/1810.08462)
+        (https://arxiv.org/abs/1810.08462).
 
     Args:
         in_channels (int): The number of bands of the input images.
         num_classes (int): The number of target classes.
-        use_dropout (bool, optional): A bool value that indicates whether to use dropout layers. 
+        use_dropout (bool, optional): A bool value that indicates whether to use dropout layers. When the model is trained 
-            When the model is trained on a relatively small dataset, the dropout layers help prevent 
+            on a relatively small dataset, the dropout layers help prevent overfitting. Default: False.
-            overfitting. Default: False.
     """
+    
     def __init__(
         self, 
         in_channels, 

paddlers/models/cd/models/unet_siamdiff.py

@@ -27,15 +27,15 @@ class UNetSiamDiff(nn.Layer):
 
     The original article refers to
         Caye Daudt, R., et al. "Fully convolutional siamese networks for change detection"
-    (https://arxiv.org/abs/1810.08462)
+        (https://arxiv.org/abs/1810.08462).
 
     Args:
         in_channels (int): The number of bands of the input images.
         num_classes (int): The number of target classes.
-        use_dropout (bool, optional): A bool value that indicates whether to use dropout layers. 
+        use_dropout (bool, optional): A bool value that indicates whether to use dropout layers. When the model is trained 
-            When the model is trained on a relatively small dataset, the dropout layers help prevent 
+            on a relatively small dataset, the dropout layers help prevent overfitting. Default: False.
-            overfitting. Default: False.
     """
+    
     def __init__(
         self, 
         in_channels,