scene 数据集

• scene数据集地址https://mulan.sourceforge.net/datasets-mlc.html

import torch
from torch.utils.data import Dataset
from scipy.io import arff
class SceneDataset(Dataset):
    """
    [scene_num, scene_feature]
        train: [1211, 294]
        test: [1196, 294]
    label
        [6,]
    """
    # 0-5 对应的是标签的下标
    meta = {'Beach': 0, 'Sunset': 1, 'FallFoliage': 2, 'Field': 3, 'Mountain': 4, 'Urban': 5}

    def __init__(self, root_path=None, transform=None):
        self.root_path = root_path
        self.transform = transform
        self.data = self._get_data_info()

    def __getitem__(self, index):
        if self.transform is not None:
            self.data[index][0] = self.transform(self.data[index][0])
        return self.data[index][0], self.data[index][1]

    def __len__(self):
        return len(self.data)

    def _get_data_info(self):
        _data_arff, _ = arff.loadarff(self.root_path)
        _data = []
        for i in range(len(_data_arff)):
            _it = list(_data_arff[i])
            _data.append([torch.tensor(_it[:-6], dtype=torch.float), torch.tensor([int(bytes.decode(encode)) for encode in _it[-6:]], dtype=torch.float)])
        return _data