import numpy as np
# Setting up data
import cv2
from keras.preprocessing.image import img_to_array
from keras.preprocessing.image import array_to_img
from keras.utils import to_categorical
from imutils import paths
from random import randint
# Used for live predictions
import time
from PIL import ImageGrab
# Used for GUI
import tkinter
from PIL import ImageTk
from PIL import Image
# Used for Generating/Labeling Data
from shutil import copyfile
import os
from random import randint
def generateTrainingImages2():
currentNumOfData = len(sorted(list(paths.list_images("generatedData/"))))
print("[INFO] Type anything and press enter to begin...")
input()
startTime = time.time()
i = 0
while (True):
if (time.time()-startTime > 1):
print("--------Captured Data--------")
im = ImageGrab.grab()
im.save("generatedData/input" + str(i+1+currentNumOfData) + ".png")
i += 1
startTime = time.time()
def labelTrainingData2():
imagePaths = sorted(list(paths.list_images("generatedData/")))
currentNumOfLabeledData = len(sorted(list(paths.list_images("trainData2/"))))
root = tkinter.Tk()
myFrame = tkinter.LabelFrame(root, text="Unlabeled Data", labelanchor="n")
myFrame.pack()
labeledCount = 0
for i in range(len(imagePaths)):
img = Image.open(imagePaths[i])
img.thumbnail((1500, 1500), Image.ANTIALIAS)
img = ImageTk.PhotoImage(img)
panel = tkinter.Label(myFrame, image = img)
panel.image = img
panel.grid(row=0, column=0)
root.update()
label = input()
if (label != 'e'):
copyfile(imagePaths[i], "trainData2/"+label+"input"+str(labeledCount+currentNumOfLabeledData)+".png")
labeledCount += 1
os.remove(imagePaths[i])
def loadTrainingImages2():
imagePaths = sorted(list(paths.list_images("trainData2/")))
x_train = np.zeros((len(imagePaths)*2, 28, 28, 3))
j = 0
for i in range(len(imagePaths)):
# Positive Label
img = cv2.imread(imagePaths[i])
arr = img_to_array(img)
#cv2.imwrite("croppped.png", arr[58:88, 702:1215])
arr = arr[58:88, 702:1215]
card = int(imagePaths[i][imagePaths[i].find('/')+1])
if (card == 0):
arr = arr[0:30, 50:104]
elif (card == 1):
arr = arr[0:30, 109:163]
elif (card == 2):
arr = arr[0:30, 168:222]
elif (card == 3):
arr = arr[0:30, 227:281]
elif (card == 4):
arr = arr[0:30, 286:340]
elif (card == 5):
arr = arr[0:30, 345:399]
elif (card == 6):
arr = arr[0:30, 404:459]
elif (card == 7):
arr = arr[0:30, 464:518]
img = arr
img = cv2.resize(img, (28, 28))
img = img_to_array(img)
x_train[j] = img
# Negative Label
img = cv2.imread(imagePaths[i])
arr = img_to_array(img)
#cv2.imwrite("croppped.png", arr[58:88, 702:1215])
arr = arr[58:88, 702:1215]
card = int(imagePaths[i][imagePaths[i].find('/')+1])
nonPlayedCards = np.arange(8)
nonPlayedCards = nonPlayedCards.tolist()
nonPlayedCards.remove(card)
cardNotPlayed = randint(0, 6)
if (cardNotPlayed == 0):
arr = arr[0:30, 50:104]
elif (cardNotPlayed == 1):
arr = arr[0:30, 109:163]
elif (cardNotPlayed == 2):
arr = arr[0:30, 168:222]
elif (cardNotPlayed == 3):
arr = arr[0:30, 227:281]
elif (cardNotPlayed == 4):
arr = arr[0:30, 286:340]
elif (cardNotPlayed == 5):
arr = arr[0:30, 345:399]
elif (cardNotPlayed == 6):
arr = arr[0:30, 404:459]
elif (cardNotPlayed == 7):
arr = arr[0:30, 464:518]
img = arr
img = cv2.resize(img, (28, 28))
img = img_to_array(img)
x_train[j+1] = img
j += 2
y_train = np.zeros(len(x_train))
for i in range(len(y_train)):
y_train[i] = (i+1)%2
return x_train, y_train
def loadTestingImages2():
img = cv2.imread("testCNN.png")
arr = img_to_array(img)
cv2.imwrite("croppped.png", arr[88:118, 702:1215])
arr = arr[88:118, 702:1215]
cv2.imwrite("testData2/output1.png", arr[0:30, 50:104])
cv2.imwrite("testData2/output2.png", arr[0:30, 109:163])
cv2.imwrite("testData2/output3.png", arr[0:30, 168:222])
cv2.imwrite("testData2/output4.png", arr[0:30, 227:281])
cv2.imwrite("testData2/output5.png", arr[0:30, 286:340])
cv2.imwrite("testData2/output6.png", arr[0:30, 345:399])
cv2.imwrite("testData2/output7.png", arr[0:30, 404:459])
cv2.imwrite("testData2/output8.png", arr[0:30, 464:518])
