jameslawlor

import numpy as np
import random 
import matplotlib.pylab as plt
import seaborn as sns
sns.set_style('white')

iters = 100000
sim_result = np.zeros(shape=(iters,1))
for i in range(iters):

jameslawlor

def BatchGenerator(getter):
    while 1:
        for f in getter.training_images_paths:
            X_train = process_images([getter.train_path + '/' + fname for fname in [f]])
            id_ = getter.get_id(f)
            y_train = np.array(getter.find_label(id_))
            y_train = np.reshape(y_train,(1,37))
            yield (X_train, y_train)

jameslawlor

def process_images(paths):
    count = len(paths)
    arr = np.zeros(shape=(count,3,106,106))
    for c, path in enumerate(paths):
        img = plt.imread(path).T
        img = img[:,106:106*3,106:106*3] #crop 424x424 -> 212x212
        img = imresize(img,size=(106,106,3),interp="cubic").T # downsample to half res
        arr[c] = img
    return arr

jameslawlor

def ConvBlock(layers, model, filters):
    for i in range(layers): 
        model.add(ZeroPadding2D((1,1)))  # zero padding of size 1
        model.add(Convolution2D(filters, 3, 3, activation='relu'))  # 3x3 filter size 
    model.add(MaxPooling2D((2,2), strides=(2,2)))

def FCBlock(model):
    model.add(Dense(4096, activation='relu'))
    model.add(Dropout(0.5))