def test_makePredictions(self):
cn.trainNetwork(5, self.testfile, self.trainfile)
cn.makePredictions(self.testfile)
rows, cols = load_extension.getDims(self.testfile)
predimage = 'prediction.tif'
testrows, testcols = load_extension.getDims(predimage)
self.assertEqual((rows, cols), (testrows, testcols))
self.assertIs(type(testrows), int)
self.assertIs(type(testcols), int)
def test_makePredictions(self):
cn.trainNetwork(5, self.testfile, self.trainfile)
cn.makePredictions(self.testfile)
rows, cols = load_extension.getDims(self.testfile)
predimage = 'prediction.tif'
testrows, testcols = load_extension.getDims(predimage)
self.assertEqual((rows, cols), (testrows, testcols))
self.assertIs(type(testrows), int)
self.assertIs(type(testcols), int)
def get_labeled_data(filename, training_file, block_size=32):
"""Read input-array (image) and label-images and return it as list of tuples. """
rows,cols = load_extension.getDims(filename)
print rows,cols
image = np.ones((rows, cols), 'uint8')
label_image = np.ones((rows, cols), 'uint8')
# x is a dummy to use as a form of error checking will return false on error
x = load_extension.getImage(image, filename)
x = load_extension.getTraining(label_image, filename, training_file)
X = []
y = []
for i in xrange(0,rows,block_size):
for j in xrange(0,cols,block_size):
try:
X.append(image[i:i + block_size, j:j + block_size].reshape(1, block_size * block_size))
y.append(int(load_extension.getLabel(label_image[i:i + block_size, j:j + block_size], "1", "0", 0.75)))
except ValueError:
continue
X = np.array(X).astype(np.float32)
label_blocks = np.array(y).astype(np.int32)
test_blocks = X.reshape(-1, 1, block_size, block_size)
return test_blocks, label_blocks
def setUpClass(self):
# pass
mt.make()
self.testfile = 'test.tif'
self.trainfile = 'train.tif'
self.downloadfile = 'TRA_000823_1720_COLOR.JP2'
temp = np.ones((256, 256), 'uint8') * 125
self.rows, self.cols = load_extension.getDims(self.testfile)
test_image = Image.fromarray(temp)
test_image.save(self.testfile)
self.image_8bit = np.array(Image.open(self.testfile))
train_image = Image.fromarray(temp)
train_image.save(self.trainfile)
self.image_8bit = np.array(Image.open(self.trainfile))
self.dims = self.image_8bit.shape
def setUpClass(self):
# pass
mt.make()
self.testfile = 'test.tif'
self.trainfile = 'train.tif'
self.downloadfile = 'TRA_000823_1720_COLOR.JP2'
temp = np.ones((256,256),'uint8')*125
self.rows, self.cols = load_extension.getDims(self.testfile)
test_image = Image.fromarray(temp)
test_image.save(self.testfile)
self.image_8bit = np.array(Image.open(self.testfile))
train_image = Image.fromarray(temp)
train_image.save(self.trainfile)
self.image_8bit = np.array(Image.open(self.trainfile))
self.dims = self.image_8bit.shape
def get_labeled_data(filename, training_file):
"""Read input-array (image) and label-images and return it as list of tuples. """
rows, cols = load_extension.getDims(filename)
print rows, cols
image = np.ones((rows,cols),'uint8')
label_image = np.ones((rows,cols),'uint8')
# x is a dummy to use as a form of error checking will return false on error
x = load_extension.getImage(image, filename)
x = load_extension.getTraining(label_image, filename, training_file)
#Seperate Image and Label into blocks
#test_blocks,blocks = create_image_blocks(768, 393,11543,rows,cols,image)
#label_blocks, blocks = create_image_blocks(768, 393,11543,rows,cols,label_image)
test_blocks,blocks = load4d(4096, 8, 8,rows,cols,image)
label_blocks, blocks = load4d(4096, 8,8,rows,cols,label_image)
#Used to Write image blocks to folder
#or i in range(blocks):
#im = Image.fromarray(test_blocks[i][i])
#im.save(str(i) +"label.tif")
return test_blocks, label_blocks
def get_labeled_data(filename, training_file):
"""Read input-array (image) and label-images and return it as list of tuples. """
rows, cols = load_extension.getDims(filename)
print rows, cols
image = np.ones((rows, cols), 'uint8')
label_image = np.ones((rows, cols), 'uint8')
# x is a dummy to use as a form of error checking will return false on error
x = load_extension.getImage(image, filename)
x = load_extension.getTraining(label_image, filename, training_file)
#Seperate Image and Label into blocks
#test_blocks,blocks = create_image_blocks(768, 393,11543,rows,cols,image)
#label_blocks, blocks = create_image_blocks(768, 393,11543,rows,cols,label_image)
test_blocks, blocks = load4d(4096, 8, 8, rows, cols, image)
label_blocks, blocks = load4d(4096, 8, 8, rows, cols, label_image)
#Used to Write image blocks to folder
#or i in range(blocks):
#im = Image.fromarray(test_blocks[i][i])
#im.save(str(i) +"label.tif")
return test_blocks, label_blocks
def getPredictionData(inputFile, block_size=32):
#Load image using extension
rows, cols = load_extension.getDims(inputFile)
print rows, cols
image = np.ones((rows, cols), 'uint8')
# x is a dummy to use as a form of error checking will return false on error
x = load_extension.getImage(image, inputFile)
X = []
blocklist = []
for i in xrange(0,rows,block_size):
for j in xrange(0,cols,block_size):
try:
X.append(image[i:i + block_size, j:j + block_size].reshape(1, block_size * block_size))
blocklist.append(image[i:i + block_size, j:j + block_size])
except ValueError:
continue
X = np.array(X).astype(np.float32)
X = X.reshape(-1, 1, block_size, block_size)
load_extension.getImage(image, inputFile)
return X, image, blocklist
import load_extension import numpy as np from PIL import Image import Tkinter root = Tkinter.Tk() image_file = "PSP_009650_1755_RED" train_file = image_file+"_dunes.tif" image_file += ".tif" rows, cols = load_extension.getDims(image_file) ratio = max((cols)/root.winfo_screenwidth(),(rows)/root.winfo_screenheight()) size = (cols /ratio , rows / ratio) sub_rows = rows/8 sub_cols = cols/4 print sub_rows, sub_cols image = np.zeros((rows,cols),"uint8") load_extension.getImage(image,image_file) # im = Image.fromarray(image[:sub_rows, sub_cols:]) #im.thumbnail(size,Image.ANTIALIAS) #im.show() # ones = np.ones((sub_rows,sub_cols),'uint8') # image = np.multiply(ones, image[:sub_rows, :sub_cols])
def test_getImage(self):
x = np.ones(le.getDims(self.filename), 'uint8')
le.getImage(x, self.filename)
self.image_8bit = np.array(Image.open(self.filename))
assert ((self.image_8bit == x).all(), True)
def test_getDims(self):
# pil_dims = Image.open()
assert(self.dims==le.getDims(self.filename), True)
def test_getPredictionData(self):
x = cn.getPredictionData(self.testfile, block_size=32)
rows, cols = load_extension.getDims(self.testfile)
testrows, testcols = x[1].shape
self.assertEqual(np.ndim(x[0]), 4)
self.assertEqual((testrows, testcols), (rows, cols))
import load_extension
import numpy as np
import matplotlib.pyplot as plt
import Tkinter
import datetime
from PIL import Image
root = Tkinter.Tk()
start = datetime.datetime.now()
#assumes you have Ryans images in the same folder as this script
filename ="PSP_009650_1755_RED.tif"
#filename = "example.tif"
#filename = "training_image.tif"
training_file = "PSP_009650_1755_RED_dunes.tif"
rows,cols = load_extension.getDims(filename)
train_image = np.zeros((rows,cols),'uint8')
image = np.zeros((rows,cols),'uint8')
load_extension.getImage(image,filename)
newfile = filename.split(".")[0] +"_flipped."+filename.split(".")[1]
start = datetime.datetime.now()
image = image[::-1]
ones = np.ones((rows,cols),'uint8')
image = np.multiply(ones, image)
load_extension.writeImage(image,newfile)
print datetime.datetime.now() -start
import load_extension
import numpy as np
import matplotlib.pyplot as plt
import Tkinter
import datetime
from PIL import Image
root = Tkinter.Tk()
start = datetime.datetime.now()
#assumes you have Ryans images in the same folder as this script
filename = "PSP_009650_1755_RED.tif"
#filename = "example.tif"
#filename = "training_image.tif"
training_file = "PSP_009650_1755_RED_dunes.tif"
rows, cols = load_extension.getDims(filename)
train_image = np.zeros((rows, cols), 'uint8')
image = np.zeros((rows, cols), 'uint8')
load_extension.getImage(image, filename)
newfile = filename.split(".")[0] + "_flipped." + filename.split(".")[1]
start = datetime.datetime.now()
image = image[::-1]
ones = np.ones((rows, cols), 'uint8')
image = np.multiply(ones, image)
load_extension.writeImage(image, newfile)
print datetime.datetime.now() - start
import load_extension
import numpy as np
from PIL import Image
import Tkinter
root = Tkinter.Tk()
image_file = "PSP_009650_1755_RED"
train_file = image_file + "_dunes.tif"
image_file += ".tif"
rows, cols = load_extension.getDims(image_file)
ratio = max((cols) / root.winfo_screenwidth(),
(rows) / root.winfo_screenheight())
size = (cols / ratio, rows / ratio)
sub_rows = rows / 8
sub_cols = cols / 4
print sub_rows, sub_cols
image = np.zeros((rows, cols), "uint8")
load_extension.getImage(image, image_file)
# im = Image.fromarray(image[:sub_rows, sub_cols:])
#im.thumbnail(size,Image.ANTIALIAS)
#im.show()
# ones = np.ones((sub_rows,sub_cols),'uint8')
# image = np.multiply(ones, image[:sub_rows, :sub_cols])
load_extension.writeImage(image[:sub_rows][sub_cols:], "test.tif")
load_extension.getImage(image, train_file)
def test_getPredictionData(self):
x = cn.getPredictionData(self.testfile, block_size=32)
rows, cols = load_extension.getDims(self.testfile)
testrows, testcols = x[1].shape
self.assertEqual(np.ndim(x[0]), 4)
self.assertEqual((testrows, testcols), (rows, cols))