def __call__(self, img, gt):
img, gt = random_mirror(img, gt)
if self.config.train_scale_array is not None:
img, gt, scale = random_scale(img, gt,
self.config.train_scale_array)
crop_size = (self.config.image_height, self.config.image_width)
crop_pos = generate_random_crop_pos(img.shape[:2], crop_size)
if self.augment:
p_img, _ = random_crop_pad_to_shape(
normalize(img, self.img_mean, self.img_std), crop_pos,
crop_size, 0)
p_img_k, _ = random_crop_pad_to_shape(
normalize(np.array(self.data_transforms(Image.fromarray(img))),
self.img_mean, self.img_std), crop_pos, crop_size, 0)
p_img = p_img.transpose(2, 0, 1)
p_img_k = p_img_k.transpose(2, 0, 1)
extra_dict = {'img_k': p_img_k}
else:
p_img, _ = random_crop_pad_to_shape(
normalize(img, self.img_mean, self.img_std), crop_pos,
crop_size, 0)
p_img = p_img.transpose(2, 0, 1)
extra_dict = None
p_gt, _ = random_crop_pad_to_shape(gt, crop_pos, crop_size, 255)
p_gt = cv2.resize(
p_gt, (self.config.image_width // self.config.gt_down_sampling,
self.config.image_height // self.config.gt_down_sampling),
interpolation=cv2.INTER_NEAREST)
return p_img, p_gt, extra_dict
def process_image(self, img, resize=None, crop_size=None):
p_img = img
if img.shape[2] < 3:
im_b = p_img
im_g = p_img
im_r = p_img
p_img = np.concatenate((im_b, im_g, im_r), axis=2)
if resize is not None:
if isinstance(resize, float):
_size = p_img.shape[:2]
# p_img = np.array(Image.fromarray(p_img).resize((int(_size[0]*resize), int(_size[1]*resize)), Image.BILINEAR))
p_img = np.array(Image.fromarray(p_img).resize((int(_size[1]*resize), int(_size[0]*resize)), Image.BILINEAR))
elif isinstance(resize, tuple) or isinstance(resize, list):
assert len(resize) == 2
p_img = np.array(Image.fromarray(p_img).resize((int(resize[0]), int(resize[1])), Image.BILINEAR))
p_img = normalize(p_img, self.image_mean, self.image_std)
if crop_size is not None:
p_img, margin = pad_image_to_shape(p_img, crop_size, cv2.BORDER_CONSTANT, value=0)
p_img = p_img.transpose(2, 0, 1)
return p_img, margin
p_img = p_img.transpose(2, 0, 1)
return p_img
def process_image(self, img, crop_size=None):
p_img = img
if img.shape[2] < 3:
im_b = p_img
im_g = p_img
im_r = p_img
p_img = np.concatenate((im_b, im_g, im_r), axis=2)
p_img = normalize(p_img, self.image_mean, self.image_std)
if crop_size is not None:
p_img, margin = pad_image_to_shape(p_img, crop_size, cv2.BORDER_CONSTANT, value=0)
p_img = p_img.transpose(2, 0, 1)
return p_img, margin
p_img = p_img.transpose(2, 0, 1)
return p_img
def __call__(self, img, gt):
img, gt = random_mirror(img, gt)
if self.config.train_scale_array is not None:
img, gt, scale = random_scale(img, gt,
self.config.train_scale_array)
img = normalize(img, self.img_mean, self.img_std)
crop_size = (self.config.image_height, self.config.image_width)
crop_pos = generate_random_crop_pos(img.shape[:2], crop_size)
p_img, _ = random_crop_pad_to_shape(img, crop_pos, crop_size, 0)
p_gt, _ = random_crop_pad_to_shape(gt, crop_pos, crop_size, 255)
p_gt = cv2.resize(
p_gt, (self.config.image_width // self.config.gt_down_sampling,
self.config.image_height // self.config.gt_down_sampling),
interpolation=cv2.INTER_NEAREST)
p_img = p_img.transpose(2, 0, 1)
extra_dict = None
return p_img, p_gt, extra_dict