File size: 11,477 Bytes
cc6c676 3c20dc8 cc6c676 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 |
from torch.utils.data import DataLoader
from torchvision import datasets, transforms
from torch.utils.data import Dataset
import torch
from configparser import ConfigParser
import matplotlib.pyplot as plt
import os
import torch as th
from PIL import Image
import numpy as np
import random
from PIL import ImageMath
import random
def dataloader(dataset, input_size, batch_size,dim,split='train', trans=False):
#transform = transforms.Compose([transforms.Resize((input_size, input_size)), transforms.ToTensor(),
# transforms.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5))])
if dataset == 'mnist':
data_loader = DataLoader(
datasets.MNIST('data/mnist', train=True, download=True, transform=transform),
batch_size=batch_size, shuffle=True)
elif dataset == 'fashion-mnist':
data_loader = DataLoader(
datasets.FashionMNIST('data/fashion-mnist', train=True, download=True, transform=transform),
batch_size=batch_size, shuffle=True)
elif dataset == 'cifar10':
data_loader = DataLoader(
datasets.CIFAR10('data/cifar10', train=True, download=True, transform=transform),
batch_size=batch_size, shuffle=True)
elif dataset == 'svhn':
data_loader = DataLoader(
datasets.SVHN('data/svhn', split=split, download=True, transform=transform),
batch_size=batch_size, shuffle=True)
elif dataset == 'stl10':
data_loader = DataLoader(
datasets.STL10('data/stl10', split=split, download=True, transform=transform),
batch_size=batch_size, shuffle=True)
elif dataset == 'lsun-bed':
data_loader = DataLoader(
datasets.LSUN('data/lsun', classes=['bedroom_train'], transform=transform),
batch_size=batch_size, shuffle=True)
elif dataset == '4cam':
if split == 'score':
cams = ScoreDataset(root_dir=os.getcwd() + '/Images/Score-Test', dim=dim, name=split, cant_images=300) #hardcode is bad but quick
return DataLoader(cams, batch_size=batch_size, shuffle=False, num_workers=0)
if split != 'test':
cams = ImagesDataset(root_dir=os.getcwd() + '/Images/ActualDataset', dim=dim, name=split, transform=trans)
return DataLoader(cams, batch_size=batch_size, shuffle=True, num_workers=0)
else:
cams = TestingDataset(root_dir=os.getcwd() + '/Images/Input-Test', dim=dim, name=split)
return DataLoader(cams, batch_size=batch_size, shuffle=False, num_workers=0)
return data_loader
class ImagesDataset(Dataset):
"""My dataset."""
def __init__(self, root_dir, dim, name, transform):
"""
Args:
root_dir (string): Directory with all the images.
transform (callable, optional): Optional transform to be applied
on a sample.
"""
self.root_dir = root_dir
self.nCameras = 2
self.imageDim = dim
self.name = name
self.parser = ConfigParser()
self.parser.read('config.ini')
self.transform = transform
def __len__(self):
return self.parser.getint(self.name, 'total')
#oneCameRoot = self.root_dir + '\CAM1'
#return int(len([name for name in os.listdir(oneCameRoot) if os.path.isfile(os.path.join(oneCameRoot, name))])/2) #por el depth
def __getitem__(self, idx):
if th.is_tensor(idx):
idx = idx.tolist()
idx = self.parser.get(self.name, str(idx))
if self.transform:
brighness = random.uniform(0.7, 1.2)
saturation = random.uniform(0, 2)
contrast = random.uniform(0.4, 2)
gamma = random.uniform(0.7, 1.3)
hue = random.uniform(-0.3, 0.3) # 0.01
oneCameRoot = self.root_dir + '/CAM0'
# foto normal
img_name = os.path.join(oneCameRoot, "n_" + idx + ".png")
img = Image.open(img_name).convert('RGB') # .convert('L')
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
if self.transform:
img = transforms.functional.adjust_gamma(img, gamma)
img = transforms.functional.adjust_brightness(img, brighness)
img = transforms.functional.adjust_contrast(img, contrast)
img = transforms.functional.adjust_saturation(img, saturation)
img = transforms.functional.adjust_hue(img, hue)
x1 = transforms.ToTensor()(img)
x1 = (x1 * 2) - 1
# foto produndidad
img_name = os.path.join(oneCameRoot, "d_" + idx + ".png")
img = Image.open(img_name).convert('I')
img = convert_I_to_L(img)
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
x1_dep = transforms.ToTensor()(img)
x1_dep = (x1_dep * 2) - 1
oneCameRoot = self.root_dir + '/CAM1'
# foto normal
img_name = os.path.join(oneCameRoot, "n_" + idx + ".png")
img = Image.open(img_name).convert('RGB') # .convert('L')
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
if self.transform:
img = transforms.functional.adjust_gamma(img, gamma)
img = transforms.functional.adjust_brightness(img, brighness)
img = transforms.functional.adjust_contrast(img, contrast)
img = transforms.functional.adjust_saturation(img, saturation)
img = transforms.functional.adjust_hue(img, hue)
x2 = transforms.ToTensor()(img)
x2 = (x2 * 2) - 1
# foto produndidad
img_name = os.path.join(oneCameRoot, "d_" + idx + ".png")
img = Image.open(img_name).convert('I')
img = convert_I_to_L(img)
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
x2_dep = transforms.ToTensor()(img)
x2_dep = (x2_dep * 2) - 1
#random izq o derecha
if (bool(random.getrandbits(1))):
sample = {'x_im': x1, 'x_dep': x1_dep, 'y_im': x2, 'y_dep': x2_dep, 'y_': torch.ones(1, self.imageDim, self.imageDim)}
else:
sample = {'x_im': x2, 'x_dep': x2_dep, 'y_im': x1, 'y_dep': x1_dep, 'y_': torch.zeros(1, self.imageDim, self.imageDim)}
return sample
def __iter__(self):
for i in range(this.__len__()):
list.append(this.__getitem__(i))
return iter(list)
class TestingDataset(Dataset):
"""My dataset."""
def __init__(self, root_dir, dim, name):
"""
Args:
root_dir (string): Directory with all the images.
transform (callable, optional): Optional transform to be applied
on a sample.
"""
self.root_dir = root_dir
self.imageDim = dim
self.name = name
files = os.listdir(self.root_dir)
self.files = [ele for ele in files if not ele.endswith('_d.png')]
def __len__(self):
#return self.parser.getint(self.name, 'total')
#oneCameRoot = self.root_dir + '\CAM1'
#return int(len([name for name in os.listdir(self.root_dir) if os.path.isfile(os.path.join(self.root_dir, name))])/2) #por el depth
return len(self.files)
def __getitem__(self, idx):
if th.is_tensor(idx):
idx = idx.tolist()
# foto normal
img_name = os.path.join(self.root_dir, self.files[idx])
img = Image.open(img_name).convert('RGB') # .convert('L')
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
x1 = transforms.ToTensor()(img)
x1 = (x1 * 2) - 1
# foto produndidad
img_name = os.path.join(self.root_dir , self.files[idx][:-4] + "_d.png")
img = Image.open(img_name)
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
x1_dep = transforms.ToTensor()(img)
x1_dep = (x1_dep * 2) - 1
sample = {'x_im': x1, 'x_dep': x1_dep}
return sample
def __iter__(self):
for i in range(this.__len__()):
list.append(this.__getitem__(i))
return iter(list)
def show_image(t_data, grey=False):
#from numpy
t_data2 = t_data.transpose(1, 2, 0)
t_data2 = t_data2 * 255.0
t_data2 = t_data2.astype(np.uint8)
if (not grey):
outIm = Image.fromarray(t_data2, mode='RGB')
else:
t_data2 = np.squeeze(t_data2, axis=2)
outIm = Image.fromarray(t_data2, mode='L')
outIm.show()
def convert_I_to_L(img):
array = np.uint8(np.array(img) / 256) #el numero esta bien, sino genera espacios en negro en la imagen
return Image.fromarray(array)
class ScoreDataset(Dataset):
"""My dataset."""
def __init__(self, root_dir, dim, name, cant_images):
"""
Args:
root_dir (string): Directory with all the images.
transform (callable, optional): Optional transform to be applied
on a sample.
"""
self.root_dir = root_dir
self.nCameras = 2
self.imageDim = dim
self.name = name
self.size = cant_images
def __len__(self):
return self.size
def __getitem__(self, idx):
oneCameRoot = self.root_dir + '/CAM0'
idx = "{:04d}".format(idx)
# foto normal
img_name = os.path.join(oneCameRoot, "n_" + idx + ".png")
img = Image.open(img_name).convert('RGB') # .convert('L')
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
x1 = transforms.ToTensor()(img)
x1 = (x1 * 2) - 1
# foto produndidad
img_name = os.path.join(oneCameRoot, "d_" + idx + ".png")
img = Image.open(img_name).convert('I')
img = convert_I_to_L(img)
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
x1_dep = transforms.ToTensor()(img)
x1_dep = (x1_dep * 2) - 1
oneCameRoot = self.root_dir + '/CAM1'
# foto normal
img_name = os.path.join(oneCameRoot, "n_" + idx + ".png")
img = Image.open(img_name).convert('RGB') # .convert('L')
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
x2 = transforms.ToTensor()(img)
x2 = (x2 * 2) - 1
# foto produndidad
img_name = os.path.join(oneCameRoot, "d_" + idx + ".png")
img = Image.open(img_name).convert('I')
img = convert_I_to_L(img)
if (img.size[0] != self.imageDim or img.size[1] != self.imageDim):
img = img.resize((self.imageDim, self.imageDim))
x2_dep = transforms.ToTensor()(img)
x2_dep = (x2_dep * 2) - 1
sample = {'x_im': x1, 'x_dep': x1_dep, 'y_im': x2, 'y_dep': x2_dep, 'y_': torch.ones(1, self.imageDim, self.imageDim)}
return sample
def __iter__(self):
for i in range(self.__len__()):
list.append(self.__getitem__(i))
return iter(list)
|