from keras.models import Sequential, load_model
from keras.layers import Dense, Convolution2D, MaxPooling2D, UpSampling2D
from sklearn.metrics import mean_squared_error
sys.path.append(modelsPath)
from importStuffLoadData import loadData
# Save time
timeStr = time.strftime('%Y%m%d_%H%M%S')
nbClasses = 1
#date = '2016_08_19'
date = '2016_11_03'
[origImRows, origImCols, (trainImages, trainResults), (valImages, valResults), (testImages, testResults), fullTestImages] = loadData(nbClasses, date)
valData = (valImages, valResults)
os.chdir(basePath)
# TODO: KL transform
# TODO: Covariance Equalization
# imresize
imRows = 64
imCols = 64
newTrainImages = np.zeros((trainImages.shape[0], imRows, imCols, 1))
for i in range(trainImages.shape[0]):
newTrainImages[i] = imresize(trainImages[i].reshape(origImRows, origImCols), (imRows, imCols)).reshape(imRows, imCols, 1)/255.0
from importStuffLoadData import loadData
from keras.models import Sequential, load_model
from keras.layers import Dense, Activation, Convolution2D, MaxPooling2D, Flatten
# Save time
timeStr = time.strftime('%Y%m%d_%H%M%S')
nbClasses = 1
#date = '2016_08_19'
date = '2016_11_03'
[
origImRows, origImCols, origImChannels, (trainImages, trainResults),
(valImages, valResults), (testImages, testResults), fullTestImages
] = loadData(nbClasses, date)
valData = (valImages, valResults)
os.chdir(basePath)
'''
[imRows, imCols, (trainImages, trainResults), trainImagesMean, valData, (testImages, testResults), fulltestImages] = loadData(nbClasses, date)
# Find trainnImagesMean
trainImagesMean = trainImages.mean(axis=0)
# Save trainImagesMean
trainImagesMeanFileName = basePath+'trainImagesMean_'+timeStr+'.npy'
np.save(trainImagesMeanFileName, trainImagesMean)
print "saved " + trainImagesMeanFileName