import sys
import os
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import tensorflow as tf
from pickle import dump, load
from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.keras.models import load_model, Model
from tensorflow.keras.applications.vgg16 import VGG16
from tensorflow.keras.preprocessing.image import load_img, img_to_array, array_to_img
from tensorflow.keras.applications.vgg16 import decode_predictions, preprocess_input
Images_With = 224
Images_Height = 224
Images_channel = 3
Images_Size = (Images_With, Images_Height)
print('Images_With =',Images_With,'- Images_Height =',Images_Height,'- Images_channel =',Images_channel)
Images_With = 224 - Images_Height = 224 - Images_channel = 3
def vgg16_classifier_default():
model = VGG16()
return model
Return: image_org - image_rs (Resize) - image_array (ndarray) - image_rsh (Reshape)
def format_data_img(filename):
#load img orginal
image_org = plt.imread(filename)
#load img and resize
image_rs = load_img(filename, target_size = (Images_With, Images_Height))
#convert to numpy
image_array = img_to_array(image_rs)
#reshape img for the model
image_rsh = image_array.reshape((1,image_array.shape[0],image_array.shape[1],image_array.shape[2]))
return (image_rs, image_org, image_array, image_rsh)
def vgg16_classifier_predict(img, model):
#Predict the probability across all output classes
yhat = model.predict(img)
#Convert the probabilities to class labels
label = decode_predictions(yhat)
#Retrive the most likely result
label = label[0][0]
#Final predict
final_predict_classifier = '%s (%.2f%%)' % (label[1], label[2]*100)
return final_predict_classifier
def plot_result(filename, model):
#format_data_img
image_rs, image_org, image_array, image_rsh = format_data_img(filename)
#predict
final_predict_classifier = vgg16_classifier_predict(image_rsh, model)
plt.figure(figsize=(15,4))
plt.subplot(1, 2, 1)
plt.imshow(image_org)
plt.grid(False)
plt.title(filename)
plt.xlabel(str(image_org.shape))
plt.colorbar()
plt.subplot(1, 2, 2)
plt.imshow(image_rs)
plt.grid(False)
plt.title(filename)
plt.xlabel(str(image_array.shape)+ ' - final_predict= '+ final_predict_classifier)
plt.colorbar()
plt.show()
model = vgg16_classifier_default()
model.summary()
Model: "vgg16" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= input_1 (InputLayer) [(None, 224, 224, 3)] 0 _________________________________________________________________ block1_conv1 (Conv2D) (None, 224, 224, 64) 1792 _________________________________________________________________ block1_conv2 (Conv2D) (None, 224, 224, 64) 36928 _________________________________________________________________ block1_pool (MaxPooling2D) (None, 112, 112, 64) 0 _________________________________________________________________ block2_conv1 (Conv2D) (None, 112, 112, 128) 73856 _________________________________________________________________ block2_conv2 (Conv2D) (None, 112, 112, 128) 147584 _________________________________________________________________ block2_pool (MaxPooling2D) (None, 56, 56, 128) 0 _________________________________________________________________ block3_conv1 (Conv2D) (None, 56, 56, 256) 295168 _________________________________________________________________ block3_conv2 (Conv2D) (None, 56, 56, 256) 590080 _________________________________________________________________ block3_conv3 (Conv2D) (None, 56, 56, 256) 590080 _________________________________________________________________ block3_pool (MaxPooling2D) (None, 28, 28, 256) 0 _________________________________________________________________ block4_conv1 (Conv2D) (None, 28, 28, 512) 1180160 _________________________________________________________________ block4_conv2 (Conv2D) (None, 28, 28, 512) 2359808 _________________________________________________________________ block4_conv3 (Conv2D) (None, 28, 28, 512) 2359808 _________________________________________________________________ block4_pool (MaxPooling2D) (None, 14, 14, 512) 0 _________________________________________________________________ block5_conv1 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_conv2 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_conv3 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_pool (MaxPooling2D) (None, 7, 7, 512) 0 _________________________________________________________________ flatten (Flatten) (None, 25088) 0 _________________________________________________________________ fc1 (Dense) (None, 4096) 102764544 _________________________________________________________________ fc2 (Dense) (None, 4096) 16781312 _________________________________________________________________ predictions (Dense) (None, 1000) 4097000 ================================================================= Total params: 138,357,544 Trainable params: 138,357,544 Non-trainable params: 0 _________________________________________________________________
#Define location from Data test-app
folder = './data/test-app/'
#Number img in test-app
nb_img = 6
for i in range(nb_img):
filename = folder + str(i) + '.jpg'
plot_result(filename, model)
del model
def model_vgg16_extractor():
model = VGG16()
#remove the output layer
model.layers.pop()
model = Model(inputs = model.input, outputs = model.layers[-1].output)
return model
def obt_feature(filename, model):
image = format_data_img(filename)
image_rsh = image[3]
#make input confirm to VGG16 input format
image_preproc = preprocess_input(image_rsh)
features = model.predict(image_preproc)
return features
def load_data_niew_features(filename):
fid = open(filename, 'rb')
image_features = load(fid, encoding='latin1')
fid.close()
return image_features
model = model_vgg16_extractor()
model.summary()
Model: "functional_1" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= input_2 (InputLayer) [(None, 224, 224, 3)] 0 _________________________________________________________________ block1_conv1 (Conv2D) (None, 224, 224, 64) 1792 _________________________________________________________________ block1_conv2 (Conv2D) (None, 224, 224, 64) 36928 _________________________________________________________________ block1_pool (MaxPooling2D) (None, 112, 112, 64) 0 _________________________________________________________________ block2_conv1 (Conv2D) (None, 112, 112, 128) 73856 _________________________________________________________________ block2_conv2 (Conv2D) (None, 112, 112, 128) 147584 _________________________________________________________________ block2_pool (MaxPooling2D) (None, 56, 56, 128) 0 _________________________________________________________________ block3_conv1 (Conv2D) (None, 56, 56, 256) 295168 _________________________________________________________________ block3_conv2 (Conv2D) (None, 56, 56, 256) 590080 _________________________________________________________________ block3_conv3 (Conv2D) (None, 56, 56, 256) 590080 _________________________________________________________________ block3_pool (MaxPooling2D) (None, 28, 28, 256) 0 _________________________________________________________________ block4_conv1 (Conv2D) (None, 28, 28, 512) 1180160 _________________________________________________________________ block4_conv2 (Conv2D) (None, 28, 28, 512) 2359808 _________________________________________________________________ block4_conv3 (Conv2D) (None, 28, 28, 512) 2359808 _________________________________________________________________ block4_pool (MaxPooling2D) (None, 14, 14, 512) 0 _________________________________________________________________ block5_conv1 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_conv2 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_conv3 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_pool (MaxPooling2D) (None, 7, 7, 512) 0 _________________________________________________________________ flatten (Flatten) (None, 25088) 0 _________________________________________________________________ fc1 (Dense) (None, 4096) 102764544 _________________________________________________________________ fc2 (Dense) (None, 4096) 16781312 _________________________________________________________________ predictions (Dense) (None, 1000) 4097000 ================================================================= Total params: 138,357,544 Trainable params: 138,357,544 Non-trainable params: 0 _________________________________________________________________
#Define location from Data test-app & new-dataset
folder = './data/test-app/'
floder_out = './data/new-dataset/'
#create a dictionary
niew_features = dict()
#Number img in test-app
nb_img = 6
for i in range(nb_img):
filename = folder + str(i) + '.jpg'
features = obt_feature(filename, model)
name_id = 'niew.' + str(i) + '.jpg'
niew_features[name_id] = features
print('Feature Shape: ', features.shape)
#Don't use JSON - ndarray not JSON serializable
dump(niew_features, open(floder_out + 'new_features.pkl', 'wb'))
del model
#Define location from Data test-app
folder = './data/new-dataset/'
filename = folder + 'new_features.pkl'
#Load file new_features.pkl
niew_data_images = load_data_niew_features(filename)
print('Len new_data_images: ',len(niew_data_images))
#List keys dictonary
List_Keys = niew_data_images.keys()
print('Keys Dictionary: ',List_Keys)
#take first data in new_data_iamges
X_data_1 = niew_data_images['niew.0.jpg']
#convert to ndarray
X_data_1 = np.array(X_data_1)
#take first key
name = next(iter(niew_data_images))
print('Name: ', name,' - Shape X_data_1: ',X_data_1.shape)
Len new_data_images: 6 Keys Dictionary: dict_keys(['niew.0.jpg', 'niew.1.jpg', 'niew.2.jpg', 'niew.3.jpg', 'niew.4.jpg', 'niew.5.jpg']) Name: niew.0.jpg - Shape X_data_1: (1, 1000)
#Define location from Data test-app
folder = './data/new-dataset/'
filename = folder + 'data_batch_1.pkl'
data = dict()
#load 'data_batch_1.pkl'
with open(filename, "rb") as f:
data = load(f, encoding='latin1')
#List keys dictonary
List_Keys = data.keys()
print('Keys Dictionary: ',List_Keys)
#extract key data
X_train = data['data']
print("Size of the data Training Features: ", len(X_train))
#convert to ndarray
X_train = np.array(X_train)
print("The dimensions of the X_train: ", X_train.shape)
Keys Dictionary: dict_keys(['batch_label', 'labels', 'data', 'filenames']) Size of the data Training Features: 10000 The dimensions of the X_train: (10000, 3072)
def model_vgg16_adap(W, H, Ch):
# define cnn model
# load model
model = VGG16(include_top = False, input_shape = (W, H, Ch))
# mark load layers as not trainable
for layer in model.layers:
layer.trainable = False
# add new classifier layers
x = tf.keras.layers.Flatten(name = 'flatten')(model.layers[-1].output)
x = tf.keras.layers.Dense(4096, activation = 'relu', name = 'dense_1')(x)
x = tf.keras.layers.Dense(4096, activation = 'relu', name = 'dense_2')(x)
Output = tf.keras.layers.Dense(2, activation = 'softmax', name = 'softmax')(x)
# define new model
model = Model(inputs = model.inputs, outputs = Output)
return model
model = model_vgg16_adap(Images_With, Images_Height, Images_channel)
model.summary()
Model: "functional_3" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= input_3 (InputLayer) [(None, 224, 224, 3)] 0 _________________________________________________________________ block1_conv1 (Conv2D) (None, 224, 224, 64) 1792 _________________________________________________________________ block1_conv2 (Conv2D) (None, 224, 224, 64) 36928 _________________________________________________________________ block1_pool (MaxPooling2D) (None, 112, 112, 64) 0 _________________________________________________________________ block2_conv1 (Conv2D) (None, 112, 112, 128) 73856 _________________________________________________________________ block2_conv2 (Conv2D) (None, 112, 112, 128) 147584 _________________________________________________________________ block2_pool (MaxPooling2D) (None, 56, 56, 128) 0 _________________________________________________________________ block3_conv1 (Conv2D) (None, 56, 56, 256) 295168 _________________________________________________________________ block3_conv2 (Conv2D) (None, 56, 56, 256) 590080 _________________________________________________________________ block3_conv3 (Conv2D) (None, 56, 56, 256) 590080 _________________________________________________________________ block3_pool (MaxPooling2D) (None, 28, 28, 256) 0 _________________________________________________________________ block4_conv1 (Conv2D) (None, 28, 28, 512) 1180160 _________________________________________________________________ block4_conv2 (Conv2D) (None, 28, 28, 512) 2359808 _________________________________________________________________ block4_conv3 (Conv2D) (None, 28, 28, 512) 2359808 _________________________________________________________________ block4_pool (MaxPooling2D) (None, 14, 14, 512) 0 _________________________________________________________________ block5_conv1 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_conv2 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_conv3 (Conv2D) (None, 14, 14, 512) 2359808 _________________________________________________________________ block5_pool (MaxPooling2D) (None, 7, 7, 512) 0 _________________________________________________________________ flatten (Flatten) (None, 25088) 0 _________________________________________________________________ dense_1 (Dense) (None, 4096) 102764544 _________________________________________________________________ dense_2 (Dense) (None, 4096) 16781312 _________________________________________________________________ softmax (Dense) (None, 2) 8194 ================================================================= Total params: 134,268,738 Trainable params: 119,554,050 Non-trainable params: 14,714,688 _________________________________________________________________