robustdg/data/chestxray_loader.py

113 строки
4.4 KiB
Python

#Common imports
import os
import random
import copy
import numpy as np
import h5py
from PIL import Image
#Pytorch
import torch
import torch.utils.data as data_utils
from torchvision import datasets, transforms
from torchvision import datasets, transforms
#Base Class
from .data_loader import BaseDataLoader
class ChestXRay(BaseDataLoader):
def __init__(self, args, list_train_domains, root, transform=None, data_case='train', match_func=False):
super().__init__(args, list_train_domains, root, transform, data_case, match_func)
self.data, self.labels, self.domains, self.indices, self.objects = self._get_data()
def _get_data(self):
data_dir= self.root
to_pil = transforms.ToPILImage()
to_tensor = transforms.ToTensor()
# Choose subsets that should be included into the training
list_img = []
list_labels = []
list_idx= []
list_size= []
list_classes=[]
for domain in self.list_domains:
domain_imgs = torch.load(data_dir + domain + '_' + self.data_case + '_image.pt')
domain_imgs_org = torch.load(data_dir + domain + '_' + self.data_case + '_image_org.pt')
domain_labels = torch.load(data_dir + domain + '_' + self.data_case + '_label.pt')
domain_idx= list(range(len(domain_imgs)))
print('Image: ', domain_imgs.shape, ' Labels: ', domain_labels.shape)
print('Source Domain ', domain)
#Apply augmentation to only training dataset
if self.data_case == 'train':
list_img.append(domain_imgs)
else:
list_img.append(domain_imgs_org)
list_labels.append(domain_labels)
list_idx.append( domain_idx )
list_size.append(len(domain_imgs))
list_classes.append( len(torch.unique(domain_labels)) )
if self.match_func:
print('Match Function Updates')
num_classes= 2
for y_c in range(num_classes):
base_class_size=0
base_class_idx=-1
for d_idx, domain in enumerate( self.list_domains ):
class_idx= list_labels[d_idx] == y_c
curr_class_size= list_labels[d_idx][class_idx].shape[0]
if base_class_size < curr_class_size:
base_class_size= curr_class_size
base_class_idx= d_idx
self.base_domain_size += base_class_size
print('Max Class Size: ', base_class_size, ' Base Domain Idx: ', base_class_idx, ' Class Label: ', y_c )
# Stack
data_imgs = torch.cat(list_img)
data_labels = torch.cat(list_labels)
data_indices = np.array(list_idx)
data_indices= np.hstack(data_indices)
self.training_list_size = list_size
#No ground truth objects in ChestXRay, for reference we set them same as data indices
data_objects= copy.deepcopy(data_indices)
# Create domain labels
data_domains = torch.zeros(data_labels.size())
domain_start=0
for idx in range(len(self.list_domains)):
curr_domain_size= self.training_list_size[idx]
data_domains[ domain_start: domain_start+ curr_domain_size ] += idx
domain_start+= curr_domain_size
# Shuffle everything one more time
inds = np.arange(data_labels.size()[0])
np.random.shuffle(inds)
data_imgs = data_imgs[inds]
data_labels = data_labels[inds]
data_domains = data_domains[inds].long()
data_indices = data_indices[inds]
data_objects = data_objects[inds]
# Convert to onehot
out_classes= list_classes[0]
y = torch.eye(out_classes)
data_labels = y[data_labels]
# Convert to onehot
d = torch.eye(len(self.list_domains))
data_domains = d[data_domains]
# If shape (B,H,W) change it to (B,C,H,W) with C=1
if len(data_imgs.shape)==3:
data_imgs= data_imgs.unsqueeze(1)
print('Shape: Data ', data_imgs.shape, ' Labels ', data_labels.shape, ' Domains ', data_domains.shape, ' Indices ', data_indices.shape, ' Objects ', data_objects.shape)
return data_imgs, data_labels, data_domains, data_indices, data_objects