Spaces:
Configuration error
Configuration error
File size: 5,243 Bytes
1ab1a09 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 |
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import cv2
import numpy as np
from PIL import Image, ImageEnhance
from scipy.ndimage import distance_transform_edt
def normalize(im, mean, std):
im = im.astype(np.float32, copy=False) / 255.0
im -= mean
im /= std
return im
def resize(im, target_size=608, interp=cv2.INTER_LINEAR):
if isinstance(target_size, list) or isinstance(target_size, tuple):
w = target_size[0]
h = target_size[1]
else:
w = target_size
h = target_size
im = cv2.resize(im, (w, h), interpolation=interp)
return im
def resize_long(im, long_size=224, interpolation=cv2.INTER_LINEAR):
value = max(im.shape[0], im.shape[1])
scale = float(long_size) / float(value)
resized_width = int(round(im.shape[1] * scale))
resized_height = int(round(im.shape[0] * scale))
im = cv2.resize(
im, (resized_width, resized_height), interpolation=interpolation)
return im
def resize_short(im, short_size=224, interpolation=cv2.INTER_LINEAR):
value = min(im.shape[0], im.shape[1])
scale = float(short_size) / float(value)
resized_width = int(round(im.shape[1] * scale))
resized_height = int(round(im.shape[0] * scale))
im = cv2.resize(
im, (resized_width, resized_height), interpolation=interpolation)
return im
def horizontal_flip(im):
if len(im.shape) == 3:
im = im[:, ::-1, :]
elif len(im.shape) == 2:
im = im[:, ::-1]
return im
def vertical_flip(im):
if len(im.shape) == 3:
im = im[::-1, :, :]
elif len(im.shape) == 2:
im = im[::-1, :]
return im
def brightness(im, brightness_lower, brightness_upper):
brightness_delta = np.random.uniform(brightness_lower, brightness_upper)
im = ImageEnhance.Brightness(im).enhance(brightness_delta)
return im
def contrast(im, contrast_lower, contrast_upper):
contrast_delta = np.random.uniform(contrast_lower, contrast_upper)
im = ImageEnhance.Contrast(im).enhance(contrast_delta)
return im
def saturation(im, saturation_lower, saturation_upper):
saturation_delta = np.random.uniform(saturation_lower, saturation_upper)
im = ImageEnhance.Color(im).enhance(saturation_delta)
return im
def hue(im, hue_lower, hue_upper):
hue_delta = np.random.uniform(hue_lower, hue_upper)
im = np.array(im.convert('HSV'))
im[:, :, 0] = im[:, :, 0] + hue_delta
im = Image.fromarray(im, mode='HSV').convert('RGB')
return im
def sharpness(im, sharpness_lower, sharpness_upper):
sharpness_delta = np.random.uniform(sharpness_lower, sharpness_upper)
im = ImageEnhance.Sharpness(im).enhance(sharpness_delta)
return im
def rotate(im, rotate_lower, rotate_upper):
rotate_delta = np.random.uniform(rotate_lower, rotate_upper)
im = im.rotate(int(rotate_delta))
return im
def mask_to_onehot(mask, num_classes):
"""
Convert a mask (H, W) to onehot (K, H, W).
Args:
mask (np.ndarray): Label mask with shape (H, W)
num_classes (int): Number of classes.
Returns:
np.ndarray: Onehot mask with shape(K, H, W).
"""
_mask = [mask == i for i in range(num_classes)]
_mask = np.array(_mask).astype(np.uint8)
return _mask
def onehot_to_binary_edge(mask, radius):
"""
Convert a onehot mask (K, H, W) to a edge mask.
Args:
mask (np.ndarray): Onehot mask with shape (K, H, W)
radius (int|float): Radius of edge.
Returns:
np.ndarray: Edge mask with shape(H, W).
"""
if radius < 1:
raise ValueError('`radius` should be greater than or equal to 1')
num_classes = mask.shape[0]
edge = np.zeros(mask.shape[1:])
# pad borders
mask = np.pad(mask, ((0, 0), (1, 1), (1, 1)),
mode='constant',
constant_values=0)
for i in range(num_classes):
dist = distance_transform_edt(mask[i, :]) + distance_transform_edt(
1.0 - mask[i, :])
dist = dist[1:-1, 1:-1]
dist[dist > radius] = 0
edge += dist
edge = np.expand_dims(edge, axis=0)
edge = (edge > 0).astype(np.uint8)
return edge
def mask_to_binary_edge(mask, radius, num_classes):
"""
Convert a segmentic segmentation mask (H, W) to a binary edge mask(H, W).
Args:
mask (np.ndarray): Label mask with shape (H, W)
radius (int|float): Radius of edge.
num_classes (int): Number of classes.
Returns:
np.ndarray: Edge mask with shape(H, W).
"""
mask = mask.squeeze()
onehot = mask_to_onehot(mask, num_classes)
edge = onehot_to_binary_edge(onehot, radius)
return edge
|