def skelton(self):
layers = QtGui.QInputDialog.getInteger(self, '# of layers',"Enter your # of Layers \nincluding input but not output")
layers = layers[0]
size_layers = []
size_layers.extend([0]*layers)
for i in range(len(size_layers)):
size_layers[i] = QtGui.QInputDialog.getInteger(self, 'Size of layer',"Enter the size of Layer each layer")[0]
size_layers.append(1)
size_layers.insert(0,500)
self.n = nt.neuralnet(*size_layers)
self.d = ds.dataset(size_layers[0],1)
self.d.generateDataSet()
self.n.loadTrainingData(self.d.getTrainingDataset())
def __init__(self, parent=None):
super(ExampleApp, self).__init__(parent)
self.setupUi(self)
# palette = QtGui.QPalette()
# palette.setBrush(palette.Background,QtGui.QBrush(QtGui.QPixmap("waves.jpg")))
# self.setPalette(palette)
self.lineEdit.setDragEnabled(True)
self.lineEdit.setAcceptDrops(True)
self.progressBar =self.progressBar
self.Teach.clicked.connect(self.generations)
self.Open_file.clicked.connect(self.file_open)
self.Quit.clicked.connect(self.close_app)
self.n = nt.neuralnet(500,300,100,1)
self.d = ds.dataset(500,1)
self.d.generateDataSet()
self.n.loadTrainingData(self.d.getTrainingDataset())
self.setWindowIcon(QtGui.QIcon('wolf.png'))
self.setWindowTitle("OCR GUI by Bryan Moore")
self.construct.clicked.connect(self.skelton)
from ocrn import dataset as ds
from ocrn import feature as ft
from ocrn import neuralnet as nt
import numpy as np
print "Ocrn: Optical Character Recognition using Neural Network\nLatest version available at http://github.com/swvist\n"
n = nt.neuralnet(100,80,1)
print "Neural Network Initialized"
d = ds.dataset(100,1)
print "Training Data Set Initialized"
if d.generateDataSet():
print "Training Data Set Generated"
if n.loadTrainingData(d.getTrainingDataset()):
print "Training Data Set loaded"
while(True):
x = raw_input("q: quit \t t: teach \t e: test \nWhat?\t:\t")
if x == "q":
break
elif x == "t":
t = int(raw_input("How many times?\t:\t"))
#n.teach(t)
n.teachUntilConvergence(max=t)
elif x == "e":
e = raw_input("Enter input file\t:\t")
x = n.activate(ft.feature.getImageFeatureVector(e))
print "\nThere is a high probability that the image is '"+str(chr(x))+"'\n"
from ocrn import dataset as ds
from ocrn import feature as ft
from ocrn import neuralnet as nt
import numpy as np
print "Ocrn: Optical Character Recognition using Neural Network\nLatest version available at http://github.com/swvist\n"
n = nt.neuralnet(100,80,1)
print "Neural Network Initialized"
d = ds.dataset(100,1)
print "Training Data Set Initialized"
if d.generateDataSet():
print "Training Data Set Generated"
if n.loadTrainingData(d.getTrainingDataset()):
print "Training Data Set loaded"
while(True):
x = raw_input("q: quit \t t: teach \t e: test \nWhat?\t:\t")
if x == "q":
break
elif x == "t":
t = int(raw_input("How many times?\t:\t"))
n.teach(t)
elif x == "e":
e = raw_input("Enter input file\t:\t")
x = n.activate(ft.feature.getImageFeatureVector(e))
print "\nThere is a high probability that the image is '"+str(unichr(x))+"'\n"
else:
from ocrn import dataset as ds
from ocrn import feature as ft
from ocrn import neuralnet as nt
import numpy as np
print "\n \nOCR Prototype: Neural Networks w/ training data and test data \n \n"
n = nt.neuralnet(500,100,1)
print "Neural Network Initialized"
d = ds.dataset(500,1)
print "Training Data Set Initialized"
if d.generateDataSet():
print "Training Data Set Generated"
if n.loadTrainingData(d.getTrainingDataset()):
print "Training Data Set loaded"
while(True):
x = raw_input("q: quit \t t: teach \t e: test \nWhat?\t:\t")
if x == "q":
break
elif x == "t":
t = int(raw_input("How many times do you want to train your data?\t:\t"))
n.teach(t)
elif x == "e":
e = raw_input("Enter input file, make sure it is the absolute form and NOT in the string form\t:\t")
x = n.activate(ft.feature.getImageFeatureVector(e))
print "\nThe highest probability letter from that the image is '"+str(unichr(x))+"'\n"
else:
def __init__(self): """ Create a __new__ Grinder object. """ self.neural_network = nn.neuralnet(100,80,1) self.data_set = ds.dataset(100,1)
def reset(self): self.neural_network = nn.neuralnet(100,80,1) self.data_set = ds.dataset(100,1)
from ocrn import neuralnet as nn
import numpy
from pyfiglet import figlet_format as ascii_print
print "\n\n\n"
print "================================================================================"
print ascii_print(' Roaster OCRN', font='standard')
print "================================================================================"
print "\n"
print "Roaster Optical Character Recognition.\nFull-bodied, with an aroma of hazelnut. Version 0.1."
print "\n"
# Load up a neural network.
print "Butting our heads together...\n"
n = nn.neuralnet(100,80,1)
# Load up training set.
print "Pulling out our datasets...\n"
d = ds.dataset(100,1)
print "Generating... ",
if d.generateDataSet():
## Data set was successfully generated.
print "Done!",
print "\n"
print "Loading... ",
if n.loadTrainingData(d.getTrainingDataset()):
## Data set successfully loaded.
print "Done!",
from ocrn import dataset as ds
from ocrn import feature as ft
from ocrn import neuralnet as nt
import numpy as np
print "\n \nOCR Prototype: Neural Networks w/ training data and test data \n \n"
n = nt.neuralnet(500,300,200,100,50,1)
print "Neural Network Initialized"
d = ds.dataset(500,1)
print "Training Data Set Initialized"
if d.generateDataSet():
print "Training Data Set Generated"
if n.loadTrainingData(d.getTrainingDataset()):
print "Training Data Set loaded"
n.teach(10000)
e = '/home/rcf-40/bryanmoo/an4/bryanmoo/OCR_Neural_Network_V1.1/data/trainingdata/01014_2014_03_25_19_02_37_ward__e.png'
x = n.activate(ft.feature.getImageFeatureVector(e))
print x
#
#while(True):
# x = raw_input("q: quit \t t: teach \t e: test \nWhat?\t:\t")
# if x == "q":
# break
# elif x == "t":
# t = int(raw_input("How many times do you want to train your data?\t:\t"))
# n.teach(t)