def generator():
while True:
modality = np.random.randint(4,size=BATCH)
usuario = np.random.randint(50,size=BATCH)
X = np.zeros((BATCH,NUM_FRAMES,112,112,3))
Y = []
Y2=[]
for i in range(BATCH):
users = os.listdir(directory+dirl[modality[i]])
images = wsort(directory+dirl[modality[i]]+users[usuario[i]]+'/') #all the images
numImages = len(images)
imagens = np.random.randint(numImages)
indice = 0
valor = np.random.random()
if valor < 0.25:
flagFlip = 1
elif valor < 0.50 and valor >= 0.25:
flagFlip = 2
elif valor < 0.75 and valor >= 0.50:
flagFlip = 3
else:
flagFlip = 4
for j in range(imagens,imagens+128,8):
imagem=image.load_img(directory+dirl[modality[i]]+users[usuario[i]]+'/'+images[(j)%numImages],target_size=(112,112))
imagem = image.img_to_array(imagem)
# here you put your function to subtract the mean of vggface2 dataset
imga = utils.preprocess_input(imagem,version=2) #subtract the mean of vggface dataset
if flagFlip == 1:
X[i,indice,:,:,:] = np.flip(imga,axis=1)
elif flagFlip == 2:
X[i,indice,:,:,:] = image.apply_affine_transform(imga,theta=30, channel_axis=2, fill_mode='nearest',cval=0.,order=1)
elif flagFlip == 3:
X[i,indice,:,:,:] = np.flip(imga,axis=0)
else:
X[i,indice,:,:,:]=imga
indice = indice+1
sets = dirl[modality[i]].split('/')[1]
# You can train the model using Training and Development sets
if sets == 'Training':
label = readCSV(dirLabels[0]+'/'+users[usuario[i]]+'_Depression.csv')
else:
label = readCSV(dirLabels[1]+'/'+users[usuario[i]]+'_Depression.csv')
Y.append(label)
Y=np.array(Y)
yield X,Y
def loadData():
con = util.getConnection()
cf = ColumnFamily(con, 'videos')
tagCF = ColumnFamily(con, 'tag_videos_composite')
movies = util.readCSV('data/movies')
for movie in movies:
title = movie[0]
uploader = movie[1]
runtime = int(movie[2]) #convert to match column validator
tags = movie[3]
rowKey = title+":"+uploader
print "Inserting in videos: {}.".format(str(movie))
row = \
{
'title':title,
'user_name':uploader,
'runtime_in_sec':runtime,
'tags_csv': tags
}
cf.insert(rowKey, row)
print 'inserting tags: {}'.format(tags)
for tag in tags.split(','):
tagCF.insert(
tag.strip().lower(), #row-key = tag name
{
(uploader, rowKey): title #(uploader,videoId)=title
}
);
print 'finishished insertion.'
con.dispose()
def val_generator(): while True: X=np.zeros((BATCH,NUM_FRAMES,112,112,3)) Y=[] modality = np.random.randint(2,size=BATCH) usuario = np.random.randint(50,size=BATCH) for m in range(BATCH): users=os.listdir(dirDevelopment+dirDev[modality[m]]) images=wsort(dirDevelopment+dirDev[modality[m]]+users[usuario[m]]+'/') #all the images numImages=len(images) imagens = np.random.randint(numImages) indice=0 for j in range(imagens,imagens+128,8): imagem = image.load_img(dirDevelopment+dirDevelopment[modality[m]]+users[usuario[m]]+'/'+images[j%numImages],target_size=(112,112)) imagem=image.img_to_array(imagem) # here you put your function to subtract the mean of vggface2 dataset imga = utils.preprocess_input(imagem,version=2) #subtract the mean of vggface dataset X[m,indice,:,:,:]=imga indice=indice+1 label = readCSV(dirLabels+users[usuario[m]]+'_Depression.csv') Y.append(label) Y=np.array(Y) yield X,Y
def loadData():
con = util.getConnection()
cf = ColumnFamily(con, 'videos')
tagCF = ColumnFamily(con, 'tag_videos_sup')
movies = util.readCSV('data/movies')
for movie in movies:
title = movie[0]
uploader = movie[1]
runtime = int(movie[2]) #convert to match column validator
tags = movie[3]
rowKey = title+":"+uploader
print "Inserting in videos: {}.".format(str(movie))
cf.insert(
rowKey,
{
'title':title,
'user_name':uploader,
'runtime_in_sec':runtime,
'tags_csv': tags
})
for tag in tags.split(','):
print 'adding tag: {0} for movie: {1}'.format(tag, title)
tagCF.insert(
tag.strip().lower(), # row-key = tag name
{
uploader: { # level 1 nesting = uploader name
rowKey: title # level 2 nesting = videoId, value = title
}
}
);
print 'finishished insertion.'
con.dispose()
def loadData():
con = util.getConnection()
cf = ColumnFamily(con, 'videos_denorm')
tagCF = ColumnFamily(con, 'tag_videos_composite')
movies = util.readCSV('data/movies')
for movie in movies:
title = movie[0]
uploader = movie[1]
runtime = int(movie[2]) #convert to match column validator
tags = movie[3]
rowKey = title+":"+uploader
print "Inserting in videos: {}.".format(str(movie))
row = \
{
'title':title,
'user_name':uploader,
'runtime_in_sec':runtime,
}
for tag in tags.split(','):
print 'adding tag: {0} for movie: {1}'.format(tag, title)
row['tag:{}'.format(tag.strip().lower())] = tag.strip()
print 'inserting denorm: {}'.format(row)
cf.insert(rowKey, row)
print 'finishished insertion.'
con.dispose()
def loadData():
events = util.readCSV('data/events.csv')
eventsList = []
for event in events:
e = {
'user': event[0],
'page': event[1],
'event': event[2],
'element': event[3]
}
eventsList.append(e)
for i in range(2):
t = Thread(target=randomEvent, args=(i, eventsList))
t.start()
def loadData():
con = util.getConnection()
cf = ColumnFamily(con, 'videos')
movies = util.readCSV('data/movies')
for movie in movies:
title = movie[0]
uploader = movie[1]
runtime = int(movie[2]) #convert to match column validator
tags = movie[3]
rowKey = title+":"+uploader
print "Inserting in videos: {}.".format(str(movie))
cf.insert(
rowKey,
{
'title':title,
'user_name':uploader,
'runtime_in_sec':runtime,
'tags_csv': tags
})
print 'finishished insertion.'
con.dispose()
def loadData():
events = util.readCSV('data/events.csv')
eventsList = []
for event in events:
e = {
'user': event[0],
'page': event[1],
'event': event[2],
'element': event[3]
}
eventsList.append(e)
#set a user, some starttime to lookup during get
global userId
global startTime
global THREAD_DONE
userId = eventsList[0]['user']
startTime = datetime.datetime.utcnow()
for i in range(2):
THREAD_DONE[i] = False #set as running
t = Thread(target=randomEvent, args=(i, eventsList))
t.start()
#!/usr/bin/env python
# encoding: utf-8
import os
import cv2
import numpy as np
from six.moves import xrange
from util import readCSV, normalize
csvpath = './csv_file/ggo.csv'
datapath = '../data/ggo/train/'
count = 1
lines = readCSV(csvpath)
for line in lines[1:]:
series_uid = line[0]
center_z = int(np.round(eval(line[1])[2]))
num_z = int(line[-1])
#print series_uid, center_z, num_z
absolute_path = datapath + series_uid
images = normalize(np.load(absolute_path + "/image.npy")) * 255
print images.shape
masks = np.load(absolute_path + "/ggo_mask.npy").astype(np.int)
print masks.shape
if num_z <= 3:
seed_image = images[:, :, center_z]
seed_mask = masks[:, :, center_z]
np.save("./data/image/{}".format(count), seed_image)
np.save("./data/mask/{}".format(count), seed_mask)
count = count + 1
else:
for i in xrange(center_z - int(num_z / 2) + 1,
hidden1 = Dense(512,activation='relu',name='hidden1')(last_layer) hidden1 = Dropout(0.5)(hidden1) #--FC Layer hidden2 = Dense(512,activation='relu',name='hidden2')(hidden1) hidden2 = Dropout(0.5)(hidden2) #--Regression Layer out = Dense(1,activation='linear',name='classifier')(hidden2) custom_vgg_model = Model(rgb_model.input,out) custom_vgg_model.compile(loss='mse',optimizer=keras.optimizers.Adam(lr=0.0001,decay=0.0005)) custom_vgg_model.fit_generator(generator(),samples_per_epoch=1000,validation_data=val_generator(),validation_steps=10,epochs=2) #--Here you read the label Y=readCSV(dirLabelsTest+'_Depression.csv') buf=0 numberOfFrames = NUM_FRAMES # while (buf < numberOfFrames): #Insert here the directory of the images of test set imagem = image.load_img(dirTesting,target_size=(FRAME_HEIGHT,FRAME_WIDTH)) imagem = image.img_to_array(imagem) #Subtract the mean of VGGFace2 dataset #---put your function here imga = utils.preprocess_input(imagem,version=2) #here it is the mean value of VGGFace dataset X.append(imga) X = np.array(X) X = np.expand_dims(X,axis=0) prediction = custom_vgg_model.predict(X)
