def get_training_and_test_from_file(f_test, f_train):
test = []
training = []
for line in f_test:
if (line != ""):
string_list = line.split()
classification = string_list.pop()
np = numpy.array(string_list)
npfloat = np.astype(numpy.float)
image_data_obj = ImageData(classification, npfloat)
test.append(image_data_obj)
else:
print("Found one blank line")
for line in f_train:
if (line != ""):
string_list = line.split()
classification = string_list.pop()
np = numpy.array(string_list)
npfloat = np.astype(numpy.float)
image_data_obj = ImageData(classification, npfloat)
training.append(image_data_obj)
else:
print("Found one blank line")
return training, test
def __init__(self, imgPath, initImage, initValid):
#We should do error checking in main, but let's do it again
if (FileClass.check_dir(imgPath) == False):
print('ERROR: Directory \"%s\" does not exist' % IMAGE_PATH)
sys.exit(1)
#Getting all of the images in a manageable way
AllImages = [
ImageData.ImageData(os.path.join(root, name))
for root, dirs, files in os.walk(IMAGE_PATH) for name in files
]
self.ImgPath = AllImages
self.Images = [ImageData.ImageData(initImage)]
self.Validation = [ImageData.ImageData(initValid)]
def folder_processing(classification, folder_path, file):
data_list = []
for filepath in glob(folder_path + '\\**'):
image = imread(filepath)
resized_image = resize(image, (80, 80), anti_aliasing=True)
histograms_data = hog(resized_image,
orientations=8,
pixels_per_cell=(8, 8),
cells_per_block=(10, 10),
feature_vector=True)
data = ImageData(classification, histograms_data)
file.write(data.__repr__() + "\n")
data_list.append(data)
return data_list
def get_data_from_file(file):
image_data_list = []
separated_characters = []
character_data = []
count = 0
for line in file:
if (line != ""):
string_list = line.split()
classification = string_list.pop()
np = numpy.array(string_list)
npfloat = np.astype(numpy.float)
image_data_obj = ImageData(classification, npfloat)
character_data.append(image_data_obj)
image_data_list.append(image_data_obj)
else:
print("Found one blank line")
count += 1
if (count > 54):
count = 0
separated_characters.append(character_data)
character_data = []
return image_data_list, separated_characters
print("Seed was:", seed)
#Will later have user input to find where the images are
#Checking the directories
if (FileClass.check_dir(IMAGE_PATH) == False):
print('ERROR: Directory \"%s\" does not exist' % IMAGE_PATH)
sys.exit(1)
if (FileClass.check_dir(VALIDATION_PATH) == False):
print("ERROR: Directory \"%s\" does not exist" % VALIDATION_PATH)
sys.exit(1)
#Making an ImageData object for all of the regular images
AllImages = [
ImageData.ImageData(os.path.join(root, name))
for root, dirs, files in os.walk(IMAGE_PATH) for name in files
]
#Making an ImageData object for all of the labeled images
ValImages = [
ImageData.ImageData(os.path.join(root, name))
for root, dirs, files in os.walk(VALIDATION_PATH) for name in files
]
#Let's get all possible values in lists
Algos = ['FF', 'MCV', 'AC', 'FB', 'CV', 'WS', 'QS'] #Need to add floods
#Taking out grayscale: CV, MCV, FD
#Took out 'MCV', 'AC', FB, SC, CV, WS
#Quickshift(QS) takes a long time, so I'm taking it out for now.
betas = [i for i in range(0, 1000)]
rng = random.Random(seed)
print("Seed was:", seed)
#seed = SEED
#Will later have user input to find where the images are
#Checking the directories
if (FileClass.check_dir(IMAGE_PATH) == False):
print ('ERROR: Directory \"%s\" does not exist'%IMAGE_PATH)
sys.exit(1)
if(FileClass.check_dir(VALIDATION_PATH) == False):
print("ERROR: Directory \"%s\" does not exist"%VALIDATION_PATH)
sys.exit(1)
#Making an ImageData object for all of the regular images
AllImages = [ImageData.ImageData(os.path.join(root, name)) for
root, dirs, files in os.walk(IMAGE_PATH) for name in files]
#Making an ImageData object for all of the labeled images
ValImages = [ImageData.ImageData(os.path.join(root, name)) for
root, dirs, files in os.walk(VALIDATION_PATH) for name in
files]
#Let's get all possible values in lists
Algos = ['FB','SC','WS','CV','MCV','AC'] #Need to add floods
#Quickshift(QS) takes a long time, so I'm taking it out for now.
betas = [i for i in range(0,10000)]
tolerance = [float(i)/1000 for i in range(0,1000,1)]
scale = [i for i in range(0,10000)]
sigma = [float(i)/100 for i in range(0,10,1)]
rng = random.Random(seed)
print("Seed was:", seed)
#Will later have user input to find where the images are
#Checking the directories
if (FileClass.check_dir(IMAGE_PATH) == False):
print ('ERROR: Directory \"%s\" does not exist'%IMAGE_PATH)
sys.exit(1)
if(FileClass.check_dir(VALIDATION_PATH) == False):
print("ERROR: Directory \"%s\" does not exist"%VALIDATION_PATH)
sys.exit(1)
#Making an ImageData object for all of the regular images
AllImages = [ImageData.ImageData(os.path.join(root, name)) for
root, dirs, files in os.walk(IMAGE_PATH) for name in files]
#Making an ImageData object for all of the labeled images
ValImages = [ImageData.ImageData(os.path.join(root, name)) for
root, dirs, files in os.walk(VALIDATION_PATH) for name in
files]
#Let's get all possible values in lists
Algos = ['FF', 'MCV', 'AC', 'FB', 'CV', 'WS', 'QS'] #Need to add floods
#Taking out grayscale: CV, MCV, FD
#Took out 'MCV', 'AC', FB, SC, CV, WS
#Quickshift(QS) takes a long time, so I'm taking it out for now.
betas = [i for i in range(0,1000)]
tolerance = [float(i)/1000 for i in range(0,1000,1)]
scale = [i for i in range(0,1000)]