In this example, we will load the weights from a MNIST convolutional autoencoder example. We will restore the weights of the encoder part only, freeze the CONV layers, and train the FC layers to perform digits classification:

import tensorflow as tf 
import numpy as np 
import os 
from models import CAE_CNN_Encoder
SAVE_FOLDER='/tmp/cae_cnn_transfer' 
from tensorflow.examples.tutorials.mnist import input_data 
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)  
model = CAE_CNN_Encoder(latent_size = 20) 
model_in = model.input 
model_out = model.output 
labels_in = model.labels 

# Get all convs weights
list_convs = [v for v in tf.global_variables() if "conv" in v.name]

# Get fc1 and logits
list_fc_layers = [v for v in tf.global_variables() if "fc" in v.name or "logits" in v.name]

# Define the saver object to load only the conv variables
saver_load_autoencoder = tf.train.Saver(var_list=list_convs)

# Define saver object to save all the variables during training
saver = tf.train.Saver()

# Define loss for classification
with tf.name_scope("LOSS"):
    loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits_v2(logits=model_out,                             labels=labels_in))
correct_prediction = tf.equal(tf.argmax(model_out,1), tf.argmax(labels_in,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))    

# Solver configuration
with tf.name_scope("Solver"):
    train_step = tf.train.AdamOptimizer(1e-4).minimize(loss, var_list=list_fc_layers)

# Initialize variables
init = tf.global_variables_initializer()

# Avoid allocating the whole memory
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.200)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))

sess.run(init)

# Restore only the CONV weights (From AutoEncoder)
saver_load_autoencoder.restore(sess, "/tmp/cae_cnn/model.ckpt-34")

# Add some tensors to observe on tensorboad
tf.summary.image("input_image", model.image_in, 4)
tf.summary.scalar("loss", loss)

merged_summary = tf.summary.merge_all()
writer = tf.summary.FileWriter(SAVE_FOLDER)
writer.add_graph(sess.graph)

#####Train######
num_epoch = 200
batch_size = 10
for epoch in range(num_epoch):
    for i in range(int(mnist.train.num_examples / batch_size)):
        # Get batch of 50 images
        batch = mnist.train.next_batch(batch_size)

        # Dump summary
        if i % 5000 == 0:            
            # Other summaries
            s = sess.run(merged_summary, feed_dict={model_in:batch[0], labels_in:batch[1]})
            writer.add_summary(s,i)                     

        # Train actually here (Also get loss value)            
        _, val_loss, t_acc = sess.run((train_step, loss, accuracy), feed_dict={model_in:batch[0],                                         labels_in:batch[1]})                

      
    print('Epoch: %d/%d loss:%d' % (epoch, num_epoch, val_loss))
    print('Save model:', epoch)
    saver.save(sess, os.path.join(SAVE_FOLDER, "model.ckpt"), epoch)