TestAll.py

 # -*- coding: utf-8 -*-
from __future__ import division, print_function, absolute_import

import tflearn
from tflearn.layers.core import input_data, dropout, fully_connected
from tflearn.layers.conv import conv_2d, max_pool_2d
from tflearn.layers.estimator import regression
from tflearn.data_preprocessing import ImagePreprocessing
from tflearn.data_augmentation import ImageAugmentation
import scipy
import numpy as np
from numpy import ndarray
import argparse
import pandas as pd
from scipy.stats import itemfreq

from zipfile import ZipFile
from io import BytesIO

import PIL.Image
from IPython.display import display

import pickle
import csv

nwidth = 64
nheight = 64

archive_test = ZipFile("Data/test.zip", 'r')

s = (len(archive_test.namelist()[:])-1, nwidth, nheight,3)
allImage = np.zeros(s)
filenames = [""] * s[0]
for i in range(1,len(archive_test.namelist()[:])):
    filename = BytesIO(archive_test.read(archive_test.namelist()[i]))
    image = PIL.Image.open(filename)
    image = image.resize((nwidth, nheight))
    image = np.array(image)
    image = np.clip(image/255.0, 0.0, 1.0)
    allImage[i-1]=image
    filenames[i-1]=archive_test.namelist()[i]
    
pickle.dump(allImage, open( "Data/test" + '.p', "wb" ) )


test = pickle.load( open( "Data/test.p", "rb" ) )

train_label_raw = pd.read_csv('Data/labels.csv')
train_label = train_label_raw["breed"].as_matrix()
train_label = train_label.reshape(train_label.shape[0],1)
print(train_label.shape)

##We convert our labels into binary arrays to give CNN
##Thanks to:
##https://www.kaggle.com/kaggleslayer/simple-convolutional-n-network-with-tensorflow
def matrix_Bin(train_label):
    labels_bin = np.array([])

    labels_name, labels0 = np.unique(train_label, return_inverse=True)

    for _, i in enumerate(itemfreq(labels0)[:,0].astype(int)):
        labels_bin0 = np.where(labels0 == itemfreq(labels0)[:,0][i], 1.0, 0.0)
        labels_bin0 = labels_bin0.reshape(1, labels_bin0.shape[0])
        if(labels_bin.shape[0] == 0):
           labels_bin = labels_bin0
        else:
            labels_bin = np.concatenate((labels_bin, labels_bin0), axis=0)

    labels_bin = labels_bin.transpose()

    return labels_name, labels_bin

labels_name, labels_bin = matrix_Bin(train_label = train_label)
labels_cls = np.argmax(labels_bin, axis=1)

img_prep = ImagePreprocessing()
img_prep.add_featurewise_zero_center()
img_prep.add_featurewise_stdnorm()

img_aug = ImageAugmentation()
img_aug.add_random_flip_leftright()
img_aug.add_random_rotation(max_angle=25.)
img_aug.add_random_blur(sigma_max=3.)

network = input_data(shape=[None, nwidth, nheight, 3],
                     data_preprocessing=img_prep,
                     data_augmentation=img_aug)
network = conv_2d(network, 32, 3, activation='relu')
network = max_pool_2d(network, 2)
network = conv_2d(network, 64, 3, activation='relu')
network = conv_2d(network, 64, 3, activation='relu')
network = max_pool_2d(network, 2)
network = fully_connected(network, 512, activation='relu')
network = dropout(network, 0.5)
network = fully_connected(network, 512, activation='relu')
network = dropout(network, 0.5)
network = fully_connected(network, 512, activation='relu')
network = dropout(network, 0.5)
network = fully_connected(network, 512, activation='relu')
network = dropout(network, 0.5)
network = fully_connected(network, 512, activation='relu')
network = dropout(network, 0.5)
network = fully_connected(network, 120, activation='softmax')
network = regression(network, optimizer='adam',
                     loss='categorical_crossentropy',
                     learning_rate=0.001)

# Train using classifier
model = tflearn.DNN(network, tensorboard_verbose=0, checkpoint_path='dog_breed_identification.tfl.ckpt')
model.load("dog_breed_identification.tfl.ckpt-6800")

print("Prediction Started")

float_formatter = lambda x: "%.17f" % x
np.set_printoptions(formatter={'float_kind':float_formatter})

sol = np.append(["id"], labels_name)
sol = np.array([sol])
#print(sol)

for i in range(len(test)):
    prediction = model.predict([test[i]])
    pred = np.argmax(prediction, axis=1)
    itemindex = np.where(labels_cls==pred)
    #print(labels_cls[itemindex][i])
    print(str(i) + " " +filenames[i][5:-4])
    pred = np.array(["%.17f" %x for x in prediction.reshape(prediction.size)])
    pred = pred.reshape(prediction.shape)
    prediction = np.append([filenames[i][5:-4]], pred)
    prediction = np.array([prediction])
    #print(prediction)
    sol = np.concatenate((sol, prediction))
    #print(pred)
    #print(labels_name[labels_cls[itemindex][i]])

sol = np.array(sol)
#print(sol)
df = pd.DataFrame(data=sol)
df.to_csv("Data/results.csv", header=False, index=False)