model_id = get_model_id()
# Create the network
network = CNN(model_id)
#read DATA
Xeval, Yeval, network.eval_size = rim.read_Data("ASVspoof2017_V2_train_eval",
"eval_info.txt")
Xeval = network.normalize(Xeval) #Normalize eval data
# print(network.eval_size/network.batch_size)
#define placeholders -predict
network.define_predict_operations()
# Recover the parameters of the model
sess = tf.Session()
restore_variables(sess)
indx = 0
network.batch_size = 64
# Iterate through eval files and calculate the classification scores
# --read data and evaluate for batch_size 64 for all images
for i in range(network.eval_size): #how many images
Xbatch, Ybatch, indx = network.read_nxt_batch(Xeval, Yeval,
network.batch_size, indx)
network.predict_utterance(sess, Xbatch, Ybatch)
sess.close()
