Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/keras/src/layers/reshaping/up_sampling3d.py: 42%

1# Copyright 2015 The TensorFlow Authors. All Rights Reserved. 

2# 

3# Licensed under the Apache License, Version 2.0 (the "License"); 

4# you may not use this file except in compliance with the License. 

5# You may obtain a copy of the License at 

6# 

7# http://www.apache.org/licenses/LICENSE-2.0 

8# 

9# Unless required by applicable law or agreed to in writing, software 

10# distributed under the License is distributed on an "AS IS" BASIS, 

11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 

12# See the License for the specific language governing permissions and 

13# limitations under the License. 

14# ============================================================================== 

15"""Keras upsampling layer for 3D inputs.""" 

16 

17 

18import tensorflow.compat.v2 as tf 

19 

20from keras.src import backend 

21from keras.src.engine.base_layer import Layer 

22from keras.src.engine.input_spec import InputSpec 

23from keras.src.utils import conv_utils 

24 

25# isort: off 

26from tensorflow.python.util.tf_export import keras_export 

27 

28 

29@keras_export("keras.layers.UpSampling3D") 

30class UpSampling3D(Layer): 

31 """Upsampling layer for 3D inputs. 

32 

33 Repeats the 1st, 2nd and 3rd dimensions 

34 of the data by `size[0]`, `size[1]` and `size[2]` respectively. 

35 

36 Examples: 

37 

38 >>> input_shape = (2, 1, 2, 1, 3) 

39 >>> x = tf.constant(1, shape=input_shape) 

40 >>> y = tf.keras.layers.UpSampling3D(size=2)(x) 

41 >>> print(y.shape) 

42 (2, 2, 4, 2, 3) 

43 

44 Args: 

45 size: Int, or tuple of 3 integers. 

46 The upsampling factors for dim1, dim2 and dim3. 

47 data_format: A string, 

48 one of `channels_last` (default) or `channels_first`. 

49 The ordering of the dimensions in the inputs. 

50 `channels_last` corresponds to inputs with shape 

51 `(batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels)` 

52 while `channels_first` corresponds to inputs with shape 

53 `(batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)`. 

54 When unspecified, uses 

55 `image_data_format` value found in your Keras config file at 

56 `~/.keras/keras.json` (if exists) else 'channels_last'. 

57 Defaults to 'channels_last'. 

58 

59 Input shape: 

60 5D tensor with shape: 

61 - If `data_format` is `"channels_last"`: 

62 `(batch_size, dim1, dim2, dim3, channels)` 

63 - If `data_format` is `"channels_first"`: 

64 `(batch_size, channels, dim1, dim2, dim3)` 

65 

66 Output shape: 

67 5D tensor with shape: 

68 - If `data_format` is `"channels_last"`: 

69 `(batch_size, upsampled_dim1, upsampled_dim2, upsampled_dim3, 

70 channels)` 

71 - If `data_format` is `"channels_first"`: 

72 `(batch_size, channels, upsampled_dim1, upsampled_dim2, 

73 upsampled_dim3)` 

74 """ 

75 

76 def __init__(self, size=(2, 2, 2), data_format=None, **kwargs): 

77 self.data_format = conv_utils.normalize_data_format(data_format) 

78 self.size = conv_utils.normalize_tuple(size, 3, "size") 

79 self.input_spec = InputSpec(ndim=5) 

80 super().__init__(**kwargs) 

81 

82 def compute_output_shape(self, input_shape): 

83 input_shape = tf.TensorShape(input_shape).as_list() 

84 if self.data_format == "channels_first": 

85 dim1 = ( 

86 self.size[0] * input_shape[2] 

87 if input_shape[2] is not None 

88 else None 

89 ) 

90 dim2 = ( 

91 self.size[1] * input_shape[3] 

92 if input_shape[3] is not None 

93 else None 

94 ) 

95 dim3 = ( 

96 self.size[2] * input_shape[4] 

97 if input_shape[4] is not None 

98 else None 

99 ) 

100 return tf.TensorShape( 

101 [input_shape[0], input_shape[1], dim1, dim2, dim3] 

102 ) 

103 else: 

104 dim1 = ( 

105 self.size[0] * input_shape[1] 

106 if input_shape[1] is not None 

107 else None 

108 ) 

109 dim2 = ( 

110 self.size[1] * input_shape[2] 

111 if input_shape[2] is not None 

112 else None 

113 ) 

114 dim3 = ( 

115 self.size[2] * input_shape[3] 

116 if input_shape[3] is not None 

117 else None 

118 ) 

119 return tf.TensorShape( 

120 [input_shape[0], dim1, dim2, dim3, input_shape[4]] 

121 ) 

122 

123 def call(self, inputs): 

124 return backend.resize_volumes( 

125 inputs, self.size[0], self.size[1], self.size[2], self.data_format 

126 ) 

127 

128 def get_config(self): 

129 config = {"size": self.size, "data_format": self.data_format} 

130 base_config = super().get_config() 

131 return dict(list(base_config.items()) + list(config.items())) 

132