def __init__(self, imageFileName, ax):
"""Constructor."""
self.ax = ax
self.imAxes = None
self.image = mpimg.open(imageFileName)
self.figure = self.ax.get_figure()
self.dx = 0
self.dy = 0
def __init__(self, imageFileName, ax):
"""Constructor."""
self.ax = ax
self.imAxes = None
self.image = mpimg.open(imageFileName)
self.figure = self.ax.get_figure()
self.dx = 0
self.dy = 0
def __init__(self, input, type):
if type == 'base64':
base64String = re.sub('^data:image/.+;base64,', '',
input).decode('base64')
image = Image.open(cStringIO.StringIO(image_data))
self.__image = image
self.__backupImage = deepcopy(self.__image)
return
else:
raise ValueError('Cutter: This type is not supported')
def __init__(self, input, type):
if type == 'base64':
base64String = re.sub('^data:image/.+;base64,',
'', input).decode('base64')
image = Image.open(cStringIO.StringIO(image_data))
self.__image = image
self.__backupImage = deepcopy(self.__image)
return
else:
raise ValueError('Cutter: This type is not supported')
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])
objectDetectorTrans = transforms.Compose([transforms.ToTensor()])
t0 = time.time()
with torch.no_grad():
for i in range(len(filelist)):
print("current file: ", filelist[i])
#Alle filer follow the pattern resolution_className_index. We extract these using split
splits = filelist[i].split("_")
#the object detector needs the entire lowRes picture. We load that in as tensor
#of shape (1,3,W,H)
imagePIL = Image.open(os.path.join(valPath, filelist[i]))
image = objectDetectorTrans(imagePIL)
image_z = torch.zeros(1, 3, image.shape[1], image.shape[2])
image_z[0] = image
image_z = image_z.to(device)
#We make inference. Ignore everything other than "boxes"
detections = objectDetector(image_z)[0]["boxes"]
print("Box points: ", detections)
t1 = time.time()
print(t1 - t0, " sec")
t0 = t1
#looping through all suggested boxes
for detection in detections:
#we need the box points in the highRes picture. For that we need
matching.append(image_similarity(src))
for i in range(batch_size):
src1 = ori_imgs[i].mul_(255).add_(0.5).clamp_(
0, 255).cpu().numpy().transpose((0, 2, 3, 1))
src2 = new_imgs[i].mul_(255).add_(0.5).clamp_(
0, 255).cpu().numpy().transpose((0, 2, 3, 1))
ori_imgs[i] = ori_imgs[i]
new_imgsp[i] = ori_imgs[i]
ori_index, ori_max = img_match(ori_imgs, matching)
new_index, new_max = img_match(new_imgs, matching)
for i in range(batch_size):
ori_result = img_joint(ori_imgs[i],
mtimage.open(imgs_list[ori_index[i]]))
new_result = img_joint(new_imgs[i],
mtimage.open(imgs_list[new_index[i]]))
ori_result = Image.fromarray((ori_result * 255).astype(np.uint8))
new_result = Image.fromarray((new_result * 255).astype(np.uint8))
ori_result.save(
os.path.join(test_path, 'ori_img' + str(ori_max) + '.jpg'))
new_result.save(
os.path.join(test_path, 'new_img' + str(new_max) + '.jpg'))
# # dataroot = '/home/admin11/Data_test/celeba'
# dataroot = '/home/admin11/Data_test/test_1000'
# # test_path = '/home/admin11/1.my_zone/test_imgs'
# test_path = '/home/admin11/1.my_zone/test_imgs'