limit ssim input img type to fp32 and fp16

mindspore/nn/layer/image.py

@@ -110,6 +110,10 @@ def _check_input_filter_size(input_shape, param_name, filter_size, func_name):
     validator.check(param_name + " shape[2]", input_shape[2], "filter_size", filter_size, Rel.GE, func_name)
     validator.check(param_name + " shape[3]", input_shape[3], "filter_size", filter_size, Rel.GE, func_name)
 
+@constexpr
+def _check_input_dtype(input_dtype, param_name, allow_dtypes, cls_name):
+    validator.check_type_name(param_name, input_dtype, allow_dtypes, cls_name)
+
 class SSIM(Cell):
     r"""
     Returns SSIM index between img1 and img2.
@@ -160,6 +164,7 @@ class SSIM(Cell):
         self.mean = P.DepthwiseConv2dNative(channel_multiplier=1, kernel_size=filter_size)
 
     def construct(self, img1, img2):
+        _check_input_dtype(F.dtype(img1), "img1", [mstype.float32, mstype.float16], self.cls_name)
         _check_input_filter_size(F.shape(img1), "img1", self.filter_size, self.cls_name)
         P.SameTypeShape()(img1, img2)
         max_val = _convert_img_dtype_to_float32(self.max_val, self.max_val)

tests/ut/python/nn/test_ssim.py

@@ -35,16 +35,8 @@ class SSIMNet(nn.Cell):
 
 def test_compile():
     net = SSIMNet()
-    img1 = Tensor(np.random.random((8, 3, 16, 16)))
-    img2 = Tensor(np.random.random((8, 3, 16, 16)))
-    _executor.compile(net, img1, img2)
-
-
-def test_compile_grayscale():
-    max_val = 255
-    net = SSIMNet(max_val=max_val)
-    img1 = Tensor(np.random.randint(0, 256, (8, 1, 16, 16), np.uint8))
-    img2 = Tensor(np.random.randint(0, 256, (8, 1, 16, 16), np.uint8))
+    img1 = Tensor(np.random.random((8, 3, 16, 16)), mstype.float32)
+    img2 = Tensor(np.random.random((8, 3, 16, 16)), mstype.float32)
     _executor.compile(net, img1, img2)