REVA-QCAV/src/data/dataset.py
Laurent Fainsin 50d18a5b39 feat: random code I don't want to make commit messages for
Former-commit-id: b94db28e25c4ada7f69d65185198a701bb5d6bfd [formerly 2476ee5d84287e40c8fb341f569249dc8aaff3e5]
Former-commit-id: 0a4b7a1f925165172b009f8812d3083e70f10201
2022-07-12 11:18:03 +02:00

113 lines
3.4 KiB
Python

from pathlib import Path
import albumentations as A
import numpy as np
from albumentations.pytorch import ToTensorV2
from PIL import Image
from torch.utils.data import Dataset
class SyntheticDataset(Dataset):
def __init__(self, image_dir, transform):
self.images = list(Path(image_dir).glob("**/*.jpg"))
self.transform = transform
def __len__(self):
return len(self.images)
def __getitem__(self, index):
# open and convert image
image = np.array(Image.open(self.images[index]).convert("RGB"), dtype=np.uint8)
# create empty mask of same size
mask = np.zeros((image.shape[0], image.shape[1]), dtype=np.uint8)
# augment image and mask
augmentations = self.transform(image=image, mask=mask)
image = augmentations["image"]
mask = augmentations["mask"]
# convert image & mask to Tensor float in [0, 1]
post_process = A.Compose(
[
A.ToFloat(max_value=255),
ToTensorV2(),
],
)
augmentations = post_process(image=image, mask=mask)
image = augmentations["image"]
mask = augmentations["mask"]
# make sure image and mask are floats
image = image.float()
mask = mask.float()
return image, mask
class LabeledDataset(Dataset):
def __init__(self, image_dir):
self.images = list(Path(image_dir).glob("**/*.jpg"))
def __len__(self):
return len(self.images)
def __getitem__(self, index):
# open and convert image
image = np.array(Image.open(self.images[index]).convert("RGB"), dtype=np.uint8)
# open and convert mask
mask_path = self.images[index].parent.joinpath("MASK.PNG")
mask = np.array(Image.open(mask_path).convert("L"), dtype=np.uint8) // 255
# convert image & mask to Tensor float in [0, 1]
post_process = A.Compose(
[
A.SmallestMaxSize(1024),
A.ToFloat(max_value=255),
ToTensorV2(),
],
)
augmentations = post_process(image=image, mask=mask)
image = augmentations["image"]
mask = augmentations["mask"]
# make sure image and mask are floats, TODO: mettre dans le post_process, ToFloat Image only
image = image.float()
mask = mask.float()
return image, mask
class LabeledDataset2(Dataset):
def __init__(self, image_dir):
self.images = list(Path(image_dir).glob("**/*.jpg"))
def __len__(self):
return len(self.images)
def __getitem__(self, index):
# open and convert image
image = np.array(Image.open(self.images[index]).convert("RGB"), dtype=np.uint8)
# open and convert mask
mask_path = self.images[index].parent.joinpath("MASK.PNG")
mask = np.array(Image.open(mask_path).convert("L"), dtype=np.uint8) // 255
# convert image & mask to Tensor float in [0, 1]
post_process = A.Compose(
[
A.ToFloat(max_value=255),
ToTensorV2(),
],
)
augmentations = post_process(image=image, mask=mask)
image = augmentations["image"]
mask = augmentations["mask"]
# make sure image and mask are floats, TODO: mettre dans le post_process, ToFloat Image only
image = image.float()
mask = mask.float()
return image, mask