fix instance norm index error

python/paddle/fluid/layers/nn.py

@@ -3190,6 +3190,10 @@ def instance_norm(input,
         dtype = core.VarDesc.VarType.FP32
 
     input_shape = input.shape
+    if len(input.shape) < 2 or len(input.shape) > 5:
+        raise ValueError(
+            'expected 2D or 3D or 4D or 5D input (got {}D input, input shape is: {})'.
+            format(len(input.shape), input_shape))
     channel_num = input_shape[1]
 
     param_shape = [channel_num]

python/paddle/fluid/tests/unittests/test_instance_norm_op.py

@@ -15,6 +15,7 @@
 from __future__ import print_function
 import unittest
 import numpy as np
+import paddle
 import paddle.fluid.core as core
 import paddle.fluid as fluid
 from paddle.fluid.op import Operator
@@ -215,6 +216,15 @@ def test_errors(self):
             self.assertRaises(TypeError, fluid.layers.instance_norm, x2)
 
 
+class TestInstanceNormOpErrorCase1(unittest.TestCase):
+    def test_errors(self):
+        with program_guard(Program(), Program()):
+            # the first dimension of input for instance_norm must between [2d, 5d] 
+            x = fluid.layers.data(
+                name='x', shape=[3], dtype="float32", append_batch_size=False)
+            self.assertRaises(ValueError, paddle.static.nn.instance_norm, x)
+
+
 class TestElasticNormOp(unittest.TestCase):
     def init_test_case(self):
         self.epsilon = 1e-5

python/paddle/fluid/tests/unittests/test_instance_norm_op_v2.py

@@ -45,12 +45,12 @@ def error2d():
 
             def error3d():
                 x_data_4 = np.random.random(size=(2, 1, 3, 3)).astype('float32')
-                instance_norm3d = paddle.nn.BatchNorm3D(1)
+                instance_norm3d = paddle.nn.InstanceNorm3D(1)
                 instance_norm3d(fluid.dygraph.to_variable(x_data_4))
 
             def weight_bias_false():
                 x_data_4 = np.random.random(size=(2, 1, 3, 3)).astype('float32')
-                instance_norm3d = paddle.nn.BatchNorm3D(
+                instance_norm3d = paddle.nn.InstanceNorm3D(
                     1, weight_attr=False, bias_attr=False)
 
             with fluid.dygraph.guard(p):