[人工智能]Pytorch Dataset

'''
Dataset:
    提供读取数据和其标签的方式：
        - 获取每条数据和标签
        - 告诉我们总共有多少条数据
'''
from torch.utils.data import Dataset
from PIL import Image
import os


class DataSet(Dataset):
    def __init__(self, type):
        # self.path = os.path.abspath('.')
        self.root_path = os.getcwd()  # 获取当前工作目录
        self.type = type  # ants or bees
        self.paths = os.path.join(self.root_path, '数据集\\hymenoptera_data\\hymenoptera_data\\train', self.type)

    def __getitem__(self, idx):
        img_path = os.path.join(self.paths, os.listdir(self.paths)[idx])
        img = Image.open(img_path)
        label = self.type
        return img, label

    def __len__(self):
        return len(os.listdir(self.paths))


ant_dataset = DataSet('ants')
bee_dataset = DataSet('bees')
data = ant_dataset + bee_dataset

'''
Dataset:
    提供读取数据和其标签的方式：
        - 获取每条数据和标签
        - 告诉我们总共有多少条数据
'''
from torch.utils.data import Dataset
from PIL import Image
import os


class DataSet(Dataset):
    def __init__(self, type):
        # self.path = os.path.abspath('.')
        self.root_path = os.getcwd()  # 获取当前工作目录
        self.type = type  # ants or bees
        self.paths = os.path.join(self.root_path, '数据集\\练手数据集\\train', self.type)

    def __getitem__(self, idx):
        img_name = os.listdir(self.paths)[idx]
        img_path = os.path.join(self.paths, img_name)
        img = Image.open(img_path)
        # 将标签从文件中读取
        # label = self.type.split('_')[0]
        with open(os.path.join(os.path.join(self.root_path, '数据集\\练手数据集\\train'), 'bees_label', '{}.txt'.format(img_name.strip('.jpg'))), mode='rt', encoding='utf8') as f:
            label = f.read().strip()
        return img, label

    def __len__(self):
        return len(os.listdir(self.paths))


ant_dataset = DataSet('bees_image')
print(ant_dataset[0])