Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/keras/src/layers/pooling/global_average_pooling2d.py: 73%

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"""Global average pooling 2D layer.""" 

16 

17 

18from keras.src import backend 

19from keras.src.layers.pooling.base_global_pooling2d import GlobalPooling2D 

20 

21# isort: off 

22from tensorflow.python.util.tf_export import keras_export 

23 

24 

25@keras_export( 

26 "keras.layers.GlobalAveragePooling2D", "keras.layers.GlobalAvgPool2D" 

27) 

28class GlobalAveragePooling2D(GlobalPooling2D): 

29 """Global average pooling operation for spatial data. 

30 

31 Examples: 

32 

33 >>> input_shape = (2, 4, 5, 3) 

34 >>> x = tf.random.normal(input_shape) 

35 >>> y = tf.keras.layers.GlobalAveragePooling2D()(x) 

36 >>> print(y.shape) 

37 (2, 3) 

38 

39 Args: 

40 data_format: A string, 

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

42 The ordering of the dimensions in the inputs. 

43 `channels_last` corresponds to inputs with shape 

44 `(batch, height, width, channels)` while `channels_first` 

45 corresponds to inputs with shape 

46 `(batch, channels, height, width)`. 

47 It defaults to the `image_data_format` value found in your 

48 Keras config file at `~/.keras/keras.json`. 

49 If you never set it, then it will be "channels_last". 

50 keepdims: A boolean, whether to keep the spatial dimensions or not. 

51 If `keepdims` is `False` (default), the rank of the tensor is reduced 

52 for spatial dimensions. 

53 If `keepdims` is `True`, the spatial dimensions are retained with 

54 length 1. 

55 The behavior is the same as for `tf.reduce_mean` or `np.mean`. 

56 

57 Input shape: 

58 - If `data_format='channels_last'`: 

59 4D tensor with shape `(batch_size, rows, cols, channels)`. 

60 - If `data_format='channels_first'`: 

61 4D tensor with shape `(batch_size, channels, rows, cols)`. 

62 

63 Output shape: 

64 - If `keepdims`=False: 

65 2D tensor with shape `(batch_size, channels)`. 

66 - If `keepdims`=True: 

67 - If `data_format='channels_last'`: 

68 4D tensor with shape `(batch_size, 1, 1, channels)` 

69 - If `data_format='channels_first'`: 

70 4D tensor with shape `(batch_size, channels, 1, 1)` 

71 """ 

72 

73 def call(self, inputs): 

74 if self.data_format == "channels_last": 

75 return backend.mean(inputs, axis=[1, 2], keepdims=self.keepdims) 

76 else: 

77 return backend.mean(inputs, axis=[2, 3], keepdims=self.keepdims) 

78 

79 

80# Alias 

81 

82GlobalAvgPool2D = GlobalAveragePooling2D 

83