def encode(bimask, binary=False):
if len(bimask.shape) == 3:
return _mask.encode(bimask, binary=binary)
elif len(bimask.shape) == 2:
h, w = bimask.shape
return _mask.encode(bimask.reshape((h, w, 1), order='F'),
binary=binary)[0]
def encode_binary_mask(mask: np.ndarray) -> t.Text:
"""Converts a binary mask into OID challenge encoding ascii text."""
# check input mask --
if mask.dtype != np.bool:
raise ValueError(
"encode_binary_mask expects a binary mask, received dtype == %s" %
mask.dtype)
mask = np.squeeze(mask)
if len(mask.shape) != 2:
raise ValueError(
"encode_binary_mask expects a 2d mask, received shape == %s" %
mask.shape)
# convert input mask to expected COCO API input --
mask_to_encode = mask.reshape(mask.shape[0], mask.shape[1], 1)
mask_to_encode = mask_to_encode.astype(np.uint8)
mask_to_encode = np.asfortranarray(mask_to_encode)
# RLE encode mask --
encoded_mask = coco_mask.encode(mask_to_encode)[0]["counts"]
# compress and base64 encoding --
binary_str = zlib.compress(encoded_mask, zlib.Z_BEST_COMPRESSION)
base64_str = base64.b64encode(binary_str)
return base64_str
def polygons_to_bitmask(polygons: List[np.ndarray], height: int,
width: int) -> np.ndarray:
"""
Args:
polygons (list[ndarray]): each array has shape (Nx2,)
height, width (int)
Returns:
ndarray: a bool mask of shape (height, width)
"""
assert len(polygons) > 0, "COCOAPI does not support empty polygons"
rles = mask_util.frPyObjects(polygons, height, width)
if len(rles) >= 2:
import pycocotools._mask as _mask
rle = _mask.decode(rles)
for i in range(rle.shape[2] - 1):
if i == 0:
temp = rle[:, :, i]
temp = np.logical_xor(temp, rle[:, :, i + 1])
rle = temp.astype('uint8').reshape([28, 28, 1])
rle = _mask.encode(rle)
rle = mask_util.merge(rle)
else:
rle = mask_util.merge(rles)
return mask_util.decode(rle).astype(np.bool)
def convert_mask_to_format(mask):
mask_to_encode = mask.reshape(mask.shape[0], mask.shape[1], 1)
mask_to_encode = mask_to_encode.astype(np.uint8)
mask_to_encode = np.asfortranarray(mask_to_encode)
# RLE encode mask --
encoded_mask = coco_mask.encode(mask_to_encode)[0]["counts"]
# compress and base64 encoding --
binary_str = zlib.compress(encoded_mask, zlib.Z_BEST_COMPRESSION)
base64_str = base64.b64encode(binary_str)
return base64_str
def encode_binary_mask(mask, mask_id): #contour, image_shape):
"""Converts a binary mask into OID challenge encoding ascii text."""
mask = np.where(mask==mask_id, 1, 0).astype(np.bool)
# check input mask --
if mask.dtype != np.bool:
raise ValueError(f"encode_binary_mask expects a binary mask, received dtype == {mask.dtype}")
mask = np.squeeze(mask)
assert len(mask.shape) == 2
# convert input mask to expected COCO API input --
mask_to_encode = mask.reshape(mask.shape[0], mask.shape[1], 1)
mask_to_encode = mask_to_encode.astype(np.uint8)
mask_to_encode = np.asfortranarray(mask_to_encode)
# RLE encode mask --
encoded_mask = coco_mask.encode(mask_to_encode)[0]["counts"]
# compress and base64 encoding --
binary_str = zlib.compress(encoded_mask, zlib.Z_BEST_COMPRESSION)
base64_str = base64.b64encode(binary_str)
return base64_str.decode() #'ascii'
def encode(bimask):
if len(bimask.shape) == 3:
return _mask.encode(bimask)
elif len(bimask.shape) == 2:
h, w = bimask.shape
return _mask.encode(bimask.reshape((h, w, 1), order='F'))[0]
def gen_new_validation_csv(seg_file, bbox_file, output_file):
# First get "isGroupOf" information. Key is ImageID, LabelName, XMin (rounded to 2 decimals) concatenated together
bbox_group_dict = {}
with open(bbox_file, 'r') as f:
next(f) # Skip header line
for line in f:
image_id, label_name, x_min, _, _, _, is_group_of = line.strip(
).split(',')
# Rounding is necessary because the decimal precision of the bounding box coordinates differs between files...
key = image_id + '_' + label_name + '_' + str(
round(float(x_min), 2))
bbox_group_dict[key] = is_group_of
print('Generated IsGroupOf dictionary.')
f_out = open(output_file, 'w')
header = 'ImageID,LabelName,ImageWidth,ImageHeight,XMin,YMin,XMax,YMax,Score,Mask'
f_out.write(header + '\n')
counter = 0
with open(seg_file, 'r') as f:
next(f) # Skip header line
for line in f:
mask_name, image_id, label_name, _, x_min, y_min, x_max, y_max, _, _ = line.strip(
).split(',')
# Form full mask and image path. Assuming data is structured per the readme.
mask_folder_prefix = mask_name[0]
mask_path = '/home/dfan/datasets/open_images_segmentation/masks/validation/validation-masks-{}/{}'.format(
mask_folder_prefix,
mask_name) # e.g. validation-masks-a/blahblah.png
image_path = '/home/dfan/datasets/open_images_segmentation/images/validation/{}.jpg'.format(
image_id)
# Get image width and height
im = cv2.imread(image_path)
height = im.shape[0]
width = im.shape[1]
# Figure out IsGroupOf value
# Rounding is necessary because the decimal precision of the bounding box coordinates differs between files...
dict_key = image_id + '_' + label_name + '_' + str(
round(float(x_min), 2))
is_group_of = bbox_group_dict[dict_key]
# convert input mask to expected COCO API input --
mask = cv2.imread(mask_path, 0)
mask = mask.reshape(mask.shape[0], mask.shape[1], 1)
mask = mask.astype(np.uint8)
mask = np.asfortranarray(mask)
# RLE encode mask
encoded_mask = str(coco_mask.encode(mask)[0]["counts"], 'utf-8')
# Write ImageID,LabelName,ImageWidth,ImageHeight,XMin,YMin,XMax,YMax,IsGroupOf,Mask
f_out.write(image_id + ',' + label_name + ',' + str(width) + ',' +
str(height) + ',' + x_min + ',' + y_min + ',' + x_max +
',' + y_max + ',' + is_group_of + ',' + encoded_mask +
'\n')
counter += 1
if counter % 1000 == 0:
print('Processed {} masks.'.format(counter))
f_out.close()
print('Successfully combined validation mask information into CSV format!')
def encode(bimask):
if len(bimask.shape) == 3:
return _mask.encode(bimask)
elif len(bimask.shape) == 2:
h, w = bimask.shape
return _mask.encode(bimask.reshape((h, w, 1), order='F'))[0]
