def __init__(self, data_path, pickle_path, budget, augment=False): self.path = data_path self.augment = augment if config.stain_normalized: self.n = stainNorm_Reinhard.Normalizer() i1 = cv2.imread('./data/source.png') i1 = cv2.cvtColor(i1, cv2.COLOR_BGR2RGB) self.n.fit(i1) label = [] img_path = [] with open(pickle_path, 'rb') as f: data_budget = pickle.load(f) for i, class_name in enumerate(data_budget[budget]['patches'].keys()): img_path = img_path + [ os.path.join(class_name, x) for x in data_budget[budget]['patches'][class_name] ] num_imgs = len(data_budget[budget]['patches'][class_name]) label = label + [i] * num_imgs print(f'number of images in class {class_name} are {num_imgs}') self.imgs = img_path self.labels = label
def __init__(self, data_path, pickle_path, budget, augment=False): self.path = data_path self.augment = augment if config.stain_normalized: self.n = stainNorm_Reinhard.Normalizer() i1 = cv2.imread('./data/source.png') i1 = cv2.cvtColor(i1, cv2.COLOR_BGR2RGB) self.n.fit(i1) normal_label = [] tumour_label = [] normal_path = [] tumour_path = [] with open(pickle_path, 'rb') as f: data_budget = pickle.load(f) for image_name in data_budget[budget]['patches']['Normal']: normal_path.append(os.path.join('Normal', image_name)) normal_label.append(0) for image_name in data_budget[budget]['patches']['Tumour']: tumour_path.append(os.path.join('Tumour', image_name)) tumour_label.append(1) self.imgs = np.append(normal_path, tumour_path) self.labels = np.append(normal_label, tumour_label)