1.tf.nn.con2d()函数介绍

tf.nn.con2d()函数参数如下图所示：

tf.nn.conv2d(
? ? input,
? ? filter,
? ? strides,
? ? padding,
? ? use_cudnn_on_gpu=True,
? ? data_format='NHWC',
? ? dilations=[1, 1, 1, 1],
? ? name=None
)

参数说明参考：https://tensorflow.google.cn/api_docs/python/tf/nn/conv2d

Args
`input`	A?`Tensor`. Must be one of the following types:?`half`,?`bfloat16`,?`float32`,?`float64`. A Tensor of rank at least 4. The dimension order is interpreted according to the value of?`data_format`; with the all-but-inner-3 dimensions acting as batch dimensions. See below for details.
`filters`	A?`Tensor`. Must have the same type as?`input`. A 4-D tensor of shape?`[filter_height,?filter_width,?in_channels,?out_channels]`
`strides`	An int or list of?`ints`?that has length?`1`,?`2`?or?`4`. The stride of the sliding window for each dimension of?`input`. If a single value is given it is replicated in the?`H`?and?`W`?dimension. By default the?`N`?and?`C`?dimensions are set to 1. The dimension order is determined by the value of?`data_format`, see below for details.
`padding`	Either the?`string`?`"SAME"`?or?`"VALID"`?indicating the type of padding algorithm to use, or a list indicating the explicit paddings at the start and end of each dimension. When explicit padding is used and data_format is?`"NHWC"`, this should be in the form?`[[0,?0],?[pad_top,pad_bottom],?[pad_left,?pad_right],?[0,?0]]`. When explicit padding used and data_format is?`"NCHW"`, this should be in the form?`[[0,?0],?[0,?0],[pad_top,?pad_bottom],?[pad_left,?pad_right]]`.
`data_format`	An optional?`string`?from:?`"NHWC",?"NCHW"`. Defaults to?`"NHWC"`. Specify the data format of the input and output data. With the default format "NHWC", the data is stored in the order of:?`batch_shape + [height,?width,?channels]`. Alternatively, the format could be "NCHW", the data storage order of:?`batch_shape + [channels,?height,?width]`.
`dilations`	An int or list of?`ints`?that has length?`1`,?`2`?or?`4`, defaults to 1. The dilation factor for each dimension of`input`. If a single value is given it is replicated in the?`H`?and?`W`?dimension. By default the?`N`?and?`C`?dimensions are set to 1. If set to k > 1, there will be k-1 skipped cells between each filter element on that dimension. The dimension order is determined by the value of?`data_format`, see above for details. Dilations in the batch and depth dimensions if a 4-d tensor must be 1.
`name`	A name for the operation (optional).

2.tf.nn.con2d()卷积运算

2.1 使用数据

flower_photos数据集中选一张玫瑰花图片

2.2代码实现卷积运算

2.2.1 使用引用包

import numpy as np
import matplotlib.pyplot as plt
from PIL import Image
import tensorflow as tf

2.2.2显示读取显示图片

data = Image.open("roses_483444865_65962cea07_m.jpg")	#返回一个PIL图像对象
plt.imshow(data)
plt.show()

2.2.3对图片进行维度变换

变换为input参数所需格式，即：

[batch_size，filter_height,?filter_width,?in_channels]

这里是[1,240,180,3]

x=np.array(data)
x=x/255
x=x.reshape(1,240,180,3)
print(x.shape)

?2.2.4 数组转化为张量

image_tensor=tf.convert_to_tensor(x)
x_input = tf.cast(image_tensor,tf.float32)

?2.2.5设置filter参数

这里使用kernel_in命名，即卷积核。维度为[3,2,1,3]([filter_height,?filter_width,?in_channels,?out_channels])out_channel需要和input参数最后一个相同。

padding包含"SAME"和"VALID"，"SAME"输出结果保持一致的维度，"VALID"根据步长等参数有所变化。

kernel_in = np.array([
     [[[1, 1, 1]], [[1, 1, 1]]],
     [[[1, 1, 1]], [[1, 1, 1]]],
    [[[1, 1, 1]], [[1, 1, 1]]]])
kernel= tf.constant(kernel_in, dtype=tf.float32)
z=tf.nn.conv2d(x_input, kernel, strides=[1,1,1,1], padding='SAME')

2.2.6 结果显示

x_conv2d=np.array(z)
x_conv2d=x_conv2d.reshape(240,180,3)
#x_conv2d=np.array(x_conv2d,dtype=float)
print(x_conv2d)
plt.imshow(x_conv2d)
plt.show()

?2.2.7 不同卷积核结果显示

变换kernel

kernel_in = np.array([
     [[[-1, 0, 1]],[[-2, 0, 2]],[[-1, 0, 1]]],
     [[[-1, 0, 1]],[[-2, 0, 2]],[[-1, 0, 1]]],
    [[[-1, 0, 1]],[[-2, 0, 2]],[[-1, 0, 1]]]])

?3.完整代码

import numpy as np
import matplotlib.pyplot as plt
from PIL import Image
import tensorflow as tf

data = Image.open("roses_483444865_65962cea07_m.jpg")	#返回一个PIL图像对象
plt.imshow(data)
plt.show()

x=np.array(data)
x=x/255
x=x.reshape(1,240,180,3)


image_tensor=tf.convert_to_tensor(x)
x_input = tf.cast(image_tensor,tf.float32)

print("x_in{}",x_input.shape)

kernel_in = np.array([
     [[[-1, 0, 1]],[[-2, 0, 2]],[[-1, 0, 1]]],
     [[[-1, 0, 1]],[[-2, 0, 2]],[[-1, 0, 1]]],
    [[[-1, 0, 1]],[[-2, 0, 2]],[[-1, 0, 1]]]])
kernel = tf.constant(kernel_in, dtype=tf.float32)
z=tf.nn.conv2d(x_input, kernel, strides=[1,1,1,1], padding='SAME')

x_conv2d=np.array(z)
x_conv2d=x_conv2d.reshape(240,180,3)
plt.imshow(x_conv2d)
plt.show()