lab15_TFCnnFilter.py

#-*- coding: utf-8 -*-
#! /usr/bin/env python
'''
#------------------------------------------------------------
    filename: lan15_runCNNFilter.py

    Generation of exemplary feature of convolutional neural network
    using two types of filters: horizontal and vertical filters

    This example is originally given by
    A. Geron "Hands-On Machine Learning with Scikit-Learn and TensorFlow",
    O'REILLY 2017. page363

    written by Jaewook Kang @ Nov 2017
#------------------------------------------------------------
'''
import numpy as np
from sklearn.datasets import load_sample_image
import tensorflow as tf
import matplotlib.pyplot as plt

# load images
china   = load_sample_image("china.jpg")
flower  = load_sample_image("flower.jpg")
dataset = np.array([china,flower],dtype=np.float32)

# load images data set size
batch_size,height, width, channels = dataset.shape

# create two 7 X 7 filters
filter_size = 7
num_of_filters = 2
filters = np.zeros(shape=(filter_size,filter_size,channels,num_of_filters),dtype=np.float32)

filters[:, 3, :, 0] = 1 # vertical line filters
filters[3, :, :, 1] = 1 # horizontal line filters

# Creat a graph with with input X plus a convolutional layer
# applying the 2 filter defined above
X = tf.placeholder(tf.float32, shape=[None, height, width, channels],name='input')

stride = 2
# prediction CNN with two filters and input X
# X is the input mini-batch
# 7 X 7 filters is the set of filters to apply
# padding = 'SAME', which means the conv layer use zero padding to keep the same feature map size
# padding = 'VALID', which mean the conv layer does not use zero padding
conv_output = tf.nn.conv2d(X,filters,strides=[1,stride,stride,1],padding='SAME')


with tf.Session() as sess:
    output = sess.run(conv_output,feed_dict= {X:dataset})

hfig = plt.figure(1,figsize=(5,10))
plt.subplot(3,2,1)
plt.imshow(filters[:,:,:,0])
plt.title('Vertical filter')

plt.subplot(3,2,2)
plt.imshow(filters[:,:,:,1])
plt.title('Horizontal filter')

# plot 1st image's and feature map with vertical filter
plt.subplot(3,2,3)
plt.imshow(output[0,:,:,0], cmap='gray')

# plot 1st image's and feature map with horizontal filter
plt.subplot(3,2,4)
plt.imshow(output[0,:,:,1], cmap='gray')

# plot 2nd image's and feature map with vertical filter
plt.subplot(3,2,5)
plt.imshow(output[1,:,:,0], cmap='gray')

# plot 2nd image's and feature map with horizontal filter
plt.subplot(3,2,6)
plt.imshow(output[1,:,:,1], cmap='gray')

plt.show()