def generate_oof_masks(mask_file,
img_dir="masks",
oof_predictions_dir=None,
img_size=640,
masks_per_file=10):
os.makedirs(img_dir, exist_ok=True)
os.makedirs(os.path.join(img_dir, "original"), exist_ok=True)
os.makedirs(os.path.join(img_dir, "solution"), exist_ok=True)
mask_dir = os.path.join("/home/selim/datasets/spacenet/train_mask_binned")
pred_dir = os.path.join(oof_predictions_dir, "full_oof")
mask = cv2.imread(
os.path.join(mask_dir, mask_file[:-4].replace("RGB", "MS") + ".tif"),
cv2.IMREAD_GRAYSCALE)
mask[mask > 0] = 255
solution_mask = cv2.imread(os.path.join(pred_dir, mask_file[:-4] + ".png"),
cv2.IMREAD_GRAYSCALE)
id = mask_file[:-4]
for i in range(masks_per_file):
h_start, w_start = random.random(), random.random()
crop = random_crop(mask, img_size, img_size, h_start, w_start)
if np.sum(crop) < 2000 * 255 and random.random() < 0.9:
continue
solution_crop = random_crop(solution_mask, img_size, img_size, h_start,
w_start)
cv2.imwrite(
os.path.join(img_dir, "original", "{}_{}oof.png".format(id, i)),
crop)
cv2.imwrite(
os.path.join(img_dir, "solution", "{}_{}oof.png".format(id, i)),
solution_crop)
def test_random_crop_with_incorrectly_large_crop_size():
img = np.ones((4, 4), dtype=np.uint8)
with pytest.raises(ValueError) as exc_info:
F.random_crop(img, crop_height=8, crop_width=8, h_start=0, w_start=0)
assert str(
exc_info.value
) == 'Requested crop size (8, 8) is larger than the image size (4, 4)'
def albumentations(self, img):
img = albumentations.random_crop(img,
crop_height=64,
crop_width=64,
h_start=0,
w_start=0)
return albumentations.resize(img, height=512, width=512)
def albumentations(self, img):
img = albumentations.random_crop(img,
crop_height=64,
crop_width=64,
h_start=0,
w_start=0)
return np.ascontiguousarray(img)
def test_random_crop_float(target):
img = np.array(
[[0.01, 0.02, 0.03, 0.04], [0.05, 0.06, 0.07, 0.08], [0.09, 0.10, 0.11, 0.12], [0.13, 0.14, 0.15, 0.16]],
dtype=np.float32,
)
expected = np.array([[0.05, 0.06], [0.09, 0.10]], dtype=np.float32)
img, expected = convert_2d_to_target_format([img, expected], target=target)
cropped_img = F.random_crop(img, crop_height=2, crop_width=2, h_start=0.5, w_start=0)
assert_array_almost_equal_nulp(cropped_img, expected)
def apply(self,
img,
crop_height=0,
crop_width=0,
h_start=0,
w_start=0,
interpolation=cv2.INTER_LINEAR,
**params):
crop = F.random_crop(img, crop_height, crop_width, h_start, w_start)
return crop
def apply(self,
img,
height=1024,
width=1024,
h_start=0,
w_start=0,
interpolation=1,
**params):
img = F.random_crop(img, height, width, h_start, w_start)
return img
def test_random_crop(target):
img = np.array(
[[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12], [13, 14, 15, 16]],
dtype=np.uint8)
expected = np.array([[5, 6], [9, 10]], dtype=np.uint8)
img, expected = convert_2d_to_target_format([img, expected], target=target)
cropped_img = F.random_crop(img,
crop_height=2,
crop_width=2,
h_start=0.5,
w_start=0)
assert np.array_equal(cropped_img, expected)
def test_random_crop(target):
img = np.array(
[[1, 2, 3, 4],
[5, 6, 7, 8],
[9, 10, 11, 12],
[13, 14, 15, 16]], dtype=np.uint8)
expected = np.array(
[[5, 6],
[9, 10]], dtype=np.uint8)
if target == 'image':
img = convert_2d_to_3d(img)
expected = convert_2d_to_3d(expected)
cropped_img = F.random_crop(img, crop_height=2, crop_width=2, h_start=0.5, w_start=0)
assert np.array_equal(cropped_img, expected)
def paired_random_crop(
images: Iterable[np.ndarray],
crops_sizes: Iterable[Tuple[int, int]],
) -> Iterable[np.ndarray]:
"""Crop a random part of the input images.
Args:
images: Sequence of images.
crops_sizes: Sequence of crop sizes ``(height, width)``.
Returns:
List of crops.
"""
h_start, w_start = random.random(), random.random()
crops = [
F.random_crop(image, height, width, h_start, w_start)
for image, (height, width) in zip(images, crops_sizes)
]
return crops
def test_random_crop_with_incorrectly_large_crop_size():
img = np.ones((4, 4), dtype=np.uint8)
with pytest.raises(ValueError, message='Requested crop size (8, 8) is larger than the image size (4, 4)'):
F.random_crop(img, crop_height=8, crop_width=8, h_start=0, w_start=0)
def albumentations(self, img):
return albumentations.random_crop(img,
crop_height=64,
crop_width=64,
h_start=0,
w_start=0)
def albumentations(self, img):
return albumentations.random_crop(img, crop_height=64, crop_width=64, h_start=0, w_start=0)
def apply(self, img, h_start=0, w_start=0, **params):
h, w, _ = img.shape
return F.random_crop(img, min(self.height, h), min(self.width, w), h_start, w_start)
def apply(self, img, crop_height=0, crop_width=0, h_start=0, w_start=0, interpolation=cv2.INTER_LINEAR, **params):
h, w, _ = img.shape
crop = F.random_crop(img, min(h, crop_height), min(w, crop_width), h_start, w_start)
return F.resize(crop, self.height, self.width, interpolation)
