def get_best_images(input_folder_real, input_folder_synthetic, num_images,
output_folder):
# Count the number of images that have been saved
counter = 1
# Create output folder
if not os.path.exists(output_folder):
os.makedirs(output_folder)
# Gather synthetic image paths
synthetic_images = os.listdir(input_folder_synthetic)
random.shuffle(synthetic_images)
# Loop through synthetic images
for image in synthetic_images:
# Make sure image is compatible
if '.html' in image:
continue
# Full path of real image and synthetic image
current_image_path_real = join(input_folder_real, image)
current_image_path_fake = join(input_folder_synthetic, image)
# Load images
current_image_real = image_class(current_image_path_real)
current_image_fake = image_class(current_image_path_fake)
# Check the number of images that have been saved
if counter > num_images:
return
current_image_fake.save_image(join(output_folder, image))
counter += 1
def generate_more_images(input_folder, output_folder, num_images):
# Counter for number of images
counter = 1
os.makedirs(output_folder, exist_ok=True)
# Loop through input folder
list_dir = os.listdir(input_folder)
for (image1, image2) in product(list_dir, list_dir):
# Make sure images are different
if image1 == image2:
continue
# Check number of images saved
if counter > num_images:
return
# Read in the images
image1_load = image_class(join(input_folder, image1))
image2_load = image_class(join(input_folder, image2))
# Combine the images
result = image1_load.combine(image2_load)
# Save the result
dest_name = "{}__{}{}".format(image1, image2, counter)
imsave(join(output_folder, dest_name), result)
counter += 1
def get_stats(input_folder):
# Empty lists for necessary statistics
side_lengths = []
pixel_areas = []
sizes_bytes = []
# Loop through images
for each in os.listdir(input_folder):
# Current image
current_image_path = os.path.join(input_folder, each)
current_image = image_class(current_image_path)
# Add in the statistics
side_lengths.append(current_image.get_side_lengths()[0]) # Width
side_lengths.append(current_image.get_side_lengths()[1]) # Height
pixel_areas.append(current_image.get_area()) # Area
sizes_bytes.append(current_image.get_size()) # Size
return side_lengths, pixel_areas, sizes_bytes
sys.path.append(os.path.join(base1, 'MNIST_Load'))
sys.path.append(os.path.join(base1, 'Rozell'))
sys.path.append(os.path.join(base1, 'Image_Class'))
os.chdir(os.path.join(base1, 'MNIST_Load'))
file_path = base1 + '/Test/DB Classifier/Overnight run'
dict1_path = file_path + '/orig_dict.png'
dict2_path = file_path + '/trained_dict.png'
dict3_path = file_path + '/trained_data.csv'
write_path = file_path + '/resid_data.csv'
plot_path = file_path + '/resid_plot.png'
import image_class as ic
##Read image and convert to numpy array
im = ic.image_class(nat_path + '\\city3.jpg')
#patches = im.slice_patches()
patches = im.slice_random(50)
print (len(patches), patches[0].shape)
'''
array.flags.writeable = True
##Set RGB to zero
array[:,:,0] = 0 #R
#array[:,:,1] = 0 #G
array[:,:,2] = 0 #B
find_new_target = True #determines whether the antogonist should find a new target. Set to false after target is chosen.
new_circle = True #determines whether the antogonist should change the circle (radius/center) it is tracking
#logical to choose motion
move_in_circle = False
#Create a display surface
screen = pygame.display.set_mode(SCREEN_SIZE,0,32) #Returns a surface object (the window)
pygame.display.set_caption("Grass Adventures")
#Create background and rescale the image size
background = pygame.image.load('grass_sideview.jpg').convert()
background_size = background.get_size()
background = pygame.transform.scale(background,(int(2*background_size[0]),int(2*background_size[1])))
#Get picture of Mittens
Mittens_image = image_class.image_class('Mittens.png',0.1,[100,100])
Sunny_image = image_class.image_class('Sunny_Snow.png',0.45,[300,300])
#Create lists of flower and animal images
flower_list = []
hyacinth_image = image_class.image_class('hyacinths.png',0.1,[200,200])
lily_image = image_class.image_class('lilyplant.png',0.5,[500,500])
callalily_image = image_class.image_class('callalily.png',0.5,[700,200])
flower_list.append(hyacinth_image)
flower_list.append(lily_image)
flower_list.append(callalily_image)
animal_list = []
bunny_image = image_class.image_class('babybunny.png',0.4,[200,600])
bluebird_image = image_class.image_class('bluebird.png',0.4,[700,100])
animal_list.append(bunny_image)
fpr0=[]
fpr1=[]
fnr0=[]
fnr1=[]
sp0=[]
sp1=[]
recall0=[]
recall1=[]
recall_av=[]
presicion0=[]
presicion1=[]
presicion_av=[]
img_class=image_class(test_CTs,test_GTVs,test_Torso,test_penalize
,bunch_of_images_no=20,is_training=1,patch_window=ct_cube_size,gtv_patch_window=gtv_cube_size )
for img_indx in range(1,len(test_CTs)):
print('img_indx:%s' %(img_indx))
ss = str(test_CTs[img_indx]).split("/")
name = ss[8] + '_' + ss[9]
name = (ss[8] + '_' + ss[9] + '_' + ss[10].split('%')[0]).split('_CT')[0]
[CT_image, GTV_image, Torso_image,
volume_depth, voxel_size, origin, direction] = _rd.read_image_seg_volume(test_CTs,
test_GTVs,
test_Torso,
img_indx,
ct_cube_size,
gtv_cube_size)
dict1_path = file_path + '/orig_dict.png'
dict2_path = file_path + '/trained_dict.png'
dict3_path = file_path + '/trained_data.csv'
write_path = file_path + '/resid_data.csv'
plot_path = file_path + '/resid_plot.png'
nat_path = base1 + '/Rozell/Natural_images'
import mnist_load as mnist
import r_network_class as Lca_jack
import image_class as ic
num_rfields = 50
num_patches = 3000
im_dims = (8,8,3) #Patch shape
nat_image = ic.image_class(nat_path + '/' + 'city2.jpg')
## Get patches and set Lca variables
training_data = nat_image.slice_patches()[:num_patches]
random.shuffle(training_data)
X = np.zeros((num_patches, np.product(im_dims)))
for i in range(len(training_data)):
X[i, :] = training_data[i].flatten()
net = Lca(num_rfields, tAlpha=0.8, tLambda=1.0)
net.init(np.product(im_dims),num_rfields)
## Use my Lca class to save pre dictionary
before = np.array(np.array(net._crossbar.copy()))
d1 = Lca_jack.r_network(np.array(net._crossbar))
d1.set_dim(im_dims)
d1.save_dictionary(5, 10, dict1_path)
# Changing image dimensions
compress = args.compress
# Change image brightness
brightness = args.increase_brightness
os.makedirs(args.output_folder, exist_ok=True)
for each in os.listdir(args.input_folder):
if args.no_filter_dups is False and "dup" in each:
continue
if each.startswith(.) or each.endswith('.html'):
continue
filepath = os.path.join(args.input_folder, each)
current_image = image_class(filepath)
if compress is True:
current_image.compress(args.compression_factor)
if brightness is True:
current_image.increase_brightness(75)
if add_AtoB is True:
# For real images
# This output path is specific to our project, if you are using
# our code you should specify your own paths
output_path = os.path.join(
args.output_folder,
"AtoB_{}.jpg".format(splitext(each)[0]))
imsave(output_path, current_image.get_image())
else:
# For fake images
# This output path is specific to our project, if you are using
