def input(network,n_tfiles,n_vfiles):
# Xtrain_in,Ytrain_in : list of ndarray
network.Xtrain_in, network.Ytrain_in, network.train_size = rim.read_Data("ASVspoof2017_V2_train_fbank", "train_info.txt",n_tfiles) # Read Train data
network.mean = np.mean(network.Xtrain_in)
network.std = np.std(network.Xtrain_in)
# Normalize input train set data
# network.Xtrain_in = normalize(network.Xtrain_in)
network.Xtrain_in = mean_normalization(network.Xtrain_in,network.mean,network.std)
# read valiation data
network.Xvalid_in, network.Yvalid_in, network.dev_size = rim.read_Data("ASVspoof2017_V2_train_dev", "dev_info.txt",n_vfiles) # Read validation data
# # Normalize input validation set data
# network.Xvalid_in = normalize(network.Xvalid_in)
network.Xvalid_in = mean_normalization(network.Xvalid_in,network.mean,network.std)
def input(self):
self.Xtrain_in, self.Ytrain_in, self.train_size = rim.read_Data(
"ASVspoof2017_V2_train_fbank", "train_info.txt") # Read Train data
# print(self.Xtrain_in)
# Normalize input train set data
self.Xtrain_in = self.normalize(self.Xtrain_in)
print(self.Ytrain_in)
print("shape"+str(self.Ytrain_in.shape))
self.Xvalid_in, self.Yvalid_in, self.dev_size = rim.read_Data(
"ASVspoof2017_V2_train_dev", "dev_info.txt") # Read validation data
# Normalize input validation set data
self.Xvalid_in = self.normalize(self.Xvalid_in)
print(self.Yvalid_in)
print("shape"+str(self.Yvalid_in.shape))
import os
import tensorflow as tf
from model import CNN
from lib.model_io import get_model_id
from lib.model_io import restore_variables
import read_img as rim
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
#test reading .cmp files
import imp
imp.load_source("read_img", "/home/tassos/Desktop/Deep4Deep/src/read_img.py")
import numpy as np
import matplotlib.pyplot as plt
import read_img as rim
#rim.read_cmp_file("testRead/T_1000001.cmp")
#check read dir_name
# cmpl=rim.read_cmp_dir("ASVspoof2017_V2_train_fbank")
# print(cmpl[0],cmpl[1])
#check read labels--ok -chk onnly path of protocol_V2 dir
#cl_types= rim.read_label("train_info.txt")
#print(cl_types)
#check read data
Xdata, Υdata, nframes = rim.read_Data("ASVspoof2017_V2_train_fbank",
"train_info.txt")