def getitem_additional(self, index):
ID = self.additional_df.ID.values[index]
ppths = '/common/danylokolinko/publichpa'
X = load_RGBY_image(ppths,
'train',
ID,
channels=self.channels,
b8=self.b8)
y = list(map(int, self.additional_df.Label.values[index].split('|')))
y = one_hot_embedding(y, self.NUM_CL)
ss_mask = None
if self.segmentation:
ppth = ppths + '_mask_semantic'
ss_mask = self.load_mask_pred(ppth, ID, 'cell')
ss_mask1 = self.load_mask_pred(ppth, ID, 'nuc')
ss_mask1 = (cv2.resize(ss_mask1, ss_mask.shape[:-1]))
ss_mask = np.concatenate((ss_mask, ss_mask1), axis=-1)
if self.load_masks:
cell_mask = self.load_mask(ppths + '_mask', ID, 'cell')
if self.cell_input:
X = np.concatenate((X, np.expand_dims(cell_mask, axis=0)), axis=0)
out1 = self.transform_func(X, y, ID, cell_mask, ss_mask)
return out1
def __getitem__(self, index):
ss_mask = None
if self.additional_data_path and index >= len(self.list_IDs):
return self.getitem_additional(index - len(self.list_IDs))
ID = self.list_IDs[index]
# return np.nan, np.nan, ID, (np.nan), np.nan
X = load_RGBY_image(self.path,
'train',
ID,
channels=self.channels,
b8=self.b8)
y = list(map(int, self.labels[index].split('|')))
y = one_hot_embedding(y, self.NUM_CL)
if self.segmentation:
ss_mask = self.load_mask_pred(self.hpasegm_path, ID, 'cell')
ss_mask1 = self.load_mask_pred(self.hpasegm_path, ID, 'nuc')
ss_mask1 = (cv2.resize(ss_mask1, ss_mask.shape[:-1]))
ss_mask = np.concatenate((ss_mask, ss_mask1), axis=-1)
if self.load_masks:
cell_mask = self.load_mask(self.load_masks_path, ID, 'cell')
if self.cell_input:
X = np.concatenate((X, np.expand_dims(cell_mask, axis=0)), axis=0)
out1 = self.transform_func(X, y, ID, cell_mask, ss_mask)
return out1
def __getitem__(self, index):
image, tags = super().__getitem__(index)
if self.transform is not None:
image = self.transform(image)
label = one_hot_embedding([self.class_dic[tag] for tag in tags],
self.classes)
return image, label
def __getitem__(self, index):
image, tags, mask = super().__getitem__(index)
if self.transform is not None:
input_dic = {'image': image, 'mask': mask}
output_dic = self.transform(input_dic)
image = output_dic['image']
mask = output_dic['mask']
label = one_hot_embedding([self.class_dic[tag] for tag in tags],
self.classes)
return image, label, mask
def __getitem__(self, index):
image, image_id, tags, mask = super().__getitem__(index)
label = one_hot_embedding([self.class_dic[tag] for tag in tags],
self.classes)
return image, image_id, label, mask