Transfer Learning using Keras

Tranfer_Learning_Keras_02.py

from keras import applications
from keras.preprocessing.image import ImageDataGenerator
from keras import optimizers
from keras.models import Sequential, Model 
from keras.layers import Dropout, Flatten, Dense, GlobalAveragePooling2D
from keras import backend as k 
from keras.callbacks import ModelCheckpoint, LearningRateScheduler, TensorBoard, EarlyStopping

img_width, img_height = 256, 256

### Build the network 
img_input = Input(shape=(256, 256, 3))
x = Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv1')(img_input)
x = Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv2')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block1_pool')(x)

# Block 2
x = Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv1')(x)
x = Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv2')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block2_pool')(x)

model = Model(input = img_input, output = x)

model.summary()
"""
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_1 (InputLayer)         (None, 256, 256, 3)       0         
_________________________________________________________________
block1_conv1 (Conv2D)        (None, 256, 256, 64)      1792      
_________________________________________________________________
block1_conv2 (Conv2D)        (None, 256, 256, 64)      36928     
_________________________________________________________________
block1_pool (MaxPooling2D)   (None, 128, 128, 64)      0         
_________________________________________________________________
block2_conv1 (Conv2D)        (None, 128, 128, 128)     73856     
_________________________________________________________________
block2_conv2 (Conv2D)        (None, 128, 128, 128)     147584    
_________________________________________________________________
block2_pool (MaxPooling2D)   (None, 64, 64, 128)       0         
=================================================================
Total params: 260,160.0
Trainable params: 260,160.0
Non-trainable params: 0.0
"""

layer_dict = dict([(layer.name, layer) for layer in model.layers])
[layer.name for layer in model.layers]
"""
['input_1',
 'block1_conv1',
 'block1_conv2',
 'block1_pool',
 'block2_conv1',
 'block2_conv2',
 'block2_pool']
"""

import h5py
weights_path = 'vgg19_weights.h5' # ('https://github.com/fchollet/deep-learning-models/releases/download/v0.1/vgg19_weights_tf_dim_ordering_tf_kernels.h5)
f = h5py.File(weights_path)

list(f["model_weights"].keys())
"""
['block1_conv1',
 'block1_conv2',
 'block1_pool',
 'block2_conv1',
 'block2_conv2',
 'block2_pool',
 'block3_conv1',
 'block3_conv2',
 'block3_conv3',
 'block3_conv4',
 'block3_pool',
 'block4_conv1',
 'block4_conv2',
 'block4_conv3',
 'block4_conv4',
 'block4_pool',
 'block5_conv1',
 'block5_conv2',
 'block5_conv3',
 'block5_conv4',
 'block5_pool',
 'dense_1',
 'dense_2',
 'dense_3',
 'dropout_1',
 'global_average_pooling2d_1',
 'input_1']
"""

# list all the layer names which are in the model.
layer_names = [layer.name for layer in model.layers]


"""
# Here we are extracting model_weights for each and every layer from the .h5 file

>>> f["model_weights"]["block1_conv1"].attrs["weight_names"]
array([b'block1_conv1/kernel:0', b'block1_conv1/bias:0'], 
      dtype='|S21')
# we are assiging this array to weight_names below 

>>> f["model_weights"]["block1_conv1"]["block1_conv1/kernel:0]
<HDF5 dataset "kernel:0": shape (3, 3, 3, 64), type "<f4">
# The list comprehension (weights) stores these two weights and bias of both the layers 

>>>layer_names.index("block1_conv1")
1

>>> model.layers[1].set_weights(weights)
# This will set the weights for that particular layer.

With a for loop we can set_weights for the entire network.
"""
for i in layer_dict.keys():
    weight_names = f["model_weights"][i].attrs["weight_names"]
    weights = [f["model_weights"][i][j] for j in weight_names]
    index = layer_names.index(i)
    model.layers[index].set_weights(weights)

    
import cv2
import numpy as np
import pandas as pd
from tqdm import tqdm
import itertools
import glob

features = []
for i in tqdm(files_location):
        im = cv2.imread(i)
        im = cv2.resize(cv2.cvtColor(im, cv2.COLOR_BGR2RGB), (256, 256)).astype(np.float32) / 255.0
        im = np.expand_dims(im, axis =0)
        outcome = model_final.predict(im)
        features.append(outcome)
        
## collect these features and create a dataframe and train a classfier on top of it.

list(f["model_weights"].keys())

At this line I get a error.

KeyError: "Unable to open object (Object 'model_weights' doesn't exist)"
I am sure the path is correctly set.

I too faced the same issue and It is a strange error. I checked the complete Keras code. 'model_weights' is defined but it is wrapped up under load_weights() function. Hence I directly used load_weights() instead of using h5py.File(weights_path) and the below code worked for me.

Working code:

import h5py
weights_path = '/home/ubuntu/<<model>>_weights_tf_dim_ordering_tf_kernels.h5'
model.load_weights(weights_path, by_name=True)
layer_count=0;
for layer in model.layers:
    weights = layer.get_weights()
    layer_count=layer_count+1
    print("[INFO] Model Layer Configuration with respect to each layer weights as follows : " + str(layer.get_config()), str(layer.name),str(weights))
print("[INFO] The total number layers is : " + str(layer_count))
a.index("block1_conv1")
for i in layer_dict.keys():
    index = a.index(i)
    model.layers[index].set_weights(weights)

where a = [layer.name for layer in model.layers]

@H-Cognitum Thanks for the workaround, it saves me lot of time.

@H-Cognitum Thanks 👍

@H-Cognitum I tried to implement the solution, but how to write things for a?
It is not clear for me how to work with this: 'where a = [layer.name for layer in model.layers]'

I think it's because 'model_weights' doesn't exist in your .h5 file. You can do a simple check with:
for key in f.keys(): print(key)
So, either load .h5 with load_weights() or work around to ignore model_weights.