#!/usr/bin/env python

# coding: utf-8

# In[42]:

import tensorflow as tf

from keras.applications.resnet50 import ResNet50

#from tensorflow.keras.applications.resnet50 import ResNet50

from keras.models import Model

from keras.preprocessing import image

from keras.applications.resnet50 import preprocess_input, decode_predictions

from keras.layers import Flatten, Input

# In[39]:

from tensorflow.keras.models import Sequential

# In[3]:

filename = 'C:/Users/rania/Downloads/cifar10-pngs-in-folders/cifar10/cifar10/test/bird/0068.png'

# In[4]:

import matplotlib.pyplot as plt

import numpy as np

img = image.load_img(filename, target_size=(32, 32))

plt.imshow(img)

img = image.img_to_array(img)

img = np.expand_dims(img, axis=0)

img = preprocess_input(img)

# In[105]:

base_model = ResNet50(weights='imagenet', pooling=max, include_top = False,input_shape=(32, 32, 3))

input = Input(shape=(32,32,3),name = 'img')

x = base_model(input)

print(x.shape)

x = Flatten()(x)

print(x.shape)

model = Model(inputs=input, outputs=x)

# In[102]:

#fetched properly of lastlayer(not the softmax)

features_old = model.predict(img)

features_old

# In[60]:

base_model.summary()

# In[72]:

#trying to fetch last but fourth layer(batch normalisation)

print(base_model.layers[-3].output)

# In[97]:

from tensorflow import keras

from tensorflow.keras import layers

from tensorflow.keras.layers import InputLayer

base_model = tensorflow.keras.applications.resnet50.ResNet50(weights='imagenet', pooling=max, include_top = False)

#base_model = ResNet50(weights='imagenet', pooling=max, include_top = False)

#input = Input(shape=(32,32,3),name = 'img')

model = tensorflow.keras.Sequential()

model.add(InputLayer(input_shape=(32,32,3), name='img'))

for layer in base_model.layers[0:176]:

model.add(layer)

#intermediate_layer_model = Model(inputs=input,outputs = base_model.layers[-3].output)

# In[ ]:

features = model.predict(img)

features

# In[74]:

#from keras.models import Model

#base_model = ResNet50(weights='imagenet', pooling=max, include_top = False)

#model = base_model # include here your original model

#input = Input(shape=(32,32,3),name = 'img')

#print(input)

#layer_name = 'layer_!5'

#m = tf.keras.Sequential()

#m.add(base_model.layers[39].output)

#intermediate_layer_model = Model(inputs=input,

# outputs=model.get_layer(layer_name).output)

#intermediate_output = intermediate_layer_model.predict(data)

#model.summary()

# In[ ]:

#import tensorflow.compat.v1 as tf

#tf.disable_v2_behavior()

#intlayer = base_model.layers[100].output

#intlayer_reshape = tf.reshape(intlayer,shape = (1,1,1024))

#print(intlayer_reshape.get_shape())

#X_shape = tf.TensorShape([None]).concatenate(intlayer_reshape.get_shape()[0:])

#X = tf.placeholder_with_default(intlayer, shape=X_shape)

#X

# In[ ]: