Delete 'model/attention/CoordAttention.py'

model/attention/CoordAttention.py

@@ -1,71 +0,0 @@
-"""
-MindSpore implementation of 'CoordAttention'
-Refer to "Coordinate Attention for Efficient Mobile Network Design"
-"""
-import mindspore as ms
-from mindspore import nn
-
-
-class h_sigmoid(nn.Cell):
-    """ h sigmoid """
-    def __init__(self):
-        super().__init__()
-        self.relu = nn.ReLU6()
-
-    def construct(self, x):
-        return self.relu(x + 3) / 6
-
-
-class h_swish(nn.Cell):
-    """ h swish """
-    def __init__(self):
-        super().__init__()
-        self.sigmoid = h_sigmoid()
-
-    def construct(self, x):
-        return x * self.sigmoid(x)
-
-
-class CoordAtt(nn.Cell):
-    """ Coordinate Attention """
-    def __init__(self, inp, oup, reduction=32):
-        super().__init__()
-        self.pool_h = nn.AdaptiveAvgPool2d((None, 1))
-        self.pool_w = nn.AdaptiveAvgPool2d((1, None))
-
-        mip = max(8, inp // reduction)
-
-        self.conv1 = nn.Conv2d(inp, mip, kernel_size=1, stride=1)
-        self.bn1 = nn.BatchNorm2d(mip)
-        self.act = h_swish()
-
-        self.conv_h = nn.Conv2d(mip, oup, kernel_size=1, stride=1)
-        self.conv_w = nn.Conv2d(mip, oup, kernel_size=1, stride=1)
-
-    def construct(self, x):
-        identity = x
-
-        _, _, H, W = x.shape
-        x_h = self.pool_h(x)
-        x_w = self.pool_w(x).permute(0, 1, 3, 2)
-
-        y = ms.ops.cat([x_h, x_w], axis=2)
-        y = self.conv1(y)
-        y = self.bn1(y)
-        y = self.act(y)
-
-        x_h, x_w = ms.ops.split(y, [H, W], axis=2)
-        x_w = x_w.permute(0, 1, 3, 2)
-
-        a_h = self.conv_h(x_h).sigmoid()
-        a_w = self.conv_w(x_w).sigmoid()
-
-        out = identity * a_w * a_h
-        return out
-
-
-if __name__ == "__main__":
-    dummy_input = ms.ops.randn(50, 512, 7, 7)
-    ca = CoordAtt(512, 512)
-    output = ca(dummy_input)
-    print(output.shape)