介绍
把[batch, height, width, channels]形状的Tensor，转变成[batch，1，1， height * width * channels]
从图像角度来看，将 height 和 width 的数据 转移到 channels 维度
常用别名
- tf.space_to_depth
语法
tf.nn.space_to_depth(
input,
block_size,
data_format='NHWC',
name=None
)
参数：
- input：tensor
- block_size：block大小，一般
>= 2 - data_format：string来自的可选”NHWC”, “NCHW”, “NCHW_VECT_C”。默认为”NHWC”
- NHWC：[ batch, height, width, channels ]
- NCHW：[ batch, channels, height, width ]
- NCHW_VECT_C：未知
返回值
- tensor
例子
例子1
arr=np.array([[[[1], [2]],
[[3], [4]]]])
print(arr.shape)
tf.nn.space_to_depth(arr,2)
执行结果：
(1, 2, 2, 1)
<tf.Tensor: id=659, shape=(1, 1, 1, 4), dtype=int64, numpy=array([[[[1, 2, 3, 4]]]])>
例子2
arr=tf.reshape(tf.range(16),(1,4,4,1))
print(arr)
print("-"*50)
res=tf.nn.space_to_depth(arr,2)
print(res)
执行结果：
tf.Tensor(
[[[[ 0]
[ 1]
[ 2]
[ 3]]
[[ 4]
[ 5]
[ 6]
[ 7]]
[[ 8]
[ 9]
[10]
[11]]
[[12]
[13]
[14]
[15]]]], shape=(1, 4, 4, 1), dtype=int32)
--------------------------------------------------
tf.Tensor(
[[[[ 0 1 4 5]
[ 2 3 6 7]]
[[ 8 9 12 13]
[10 11 14 15]]]], shape=(1, 2, 2, 4), dtype=int32)
例子3
arr = np.arange(150).reshape(2,5,5,3)
print(arr.shape)
tf.nn.space_to_depth(arr,5)
执行结果：
(2, 5, 5, 3)
<tf.Tensor: id=662, shape=(2, 1, 1, 75), dtype=int64, numpy=
array([[[[ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11,
12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23,
24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,
36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47,
48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59,
60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71,
72, 73, 74]]],
[[[ 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86,
87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98,
99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110,
111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122,
123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134,
135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146,
147, 148, 149]]]])>
感谢：
https://blog.csdn.net/qq_36556893/article/details/89471553