Skip to content

svhn

load_data

Load and return the Street View House Numbers (SVHN) dataset.

Parameters:

Name Type Description Default
root_dir Optional[str]

The path to store the downloaded data. When path is not provided, the data will be saved into fastestimator_data under the user's home directory.

 None

Returns:

Type Description
Tuple[PickleDataset, PickleDataset]

(train_data, test_data)
Source code in fastestimator/fastestimator/dataset/data/svhn.py 
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
def load_data(root_dir: Optional[str] = None) -> Tuple[PickleDataset, PickleDataset]:
    """Load and return the Street View House Numbers (SVHN) dataset.

    Args:
        root_dir: The path to store the downloaded data. When `path` is not provided, the data will be saved into
            `fastestimator_data` under the user's home directory.

    Returns:
        (train_data, test_data)
    """
    home = str(Path.home())

    if root_dir is None:
        root_dir = os.path.join(home, 'fastestimator_data', 'SVHN')
    else:
        root_dir = os.path.join(os.path.abspath(root_dir), 'SVHN')
    os.makedirs(root_dir, exist_ok=True)

    train_file_path = os.path.join(root_dir, 'train.pickle')
    test_file_path = os.path.join(root_dir, 'test.pickle')
    train_compressed_path = os.path.join(root_dir, "train.tar.gz")
    test_compressed_path = os.path.join(root_dir, "test.tar.gz")
    train_folder_path = os.path.join(root_dir, "train")
    test_folder_path = os.path.join(root_dir, "test")

    if not os.path.exists(train_folder_path):
        # download
        if not os.path.exists(train_compressed_path):
            print("Downloading train data to {}".format(root_dir))
            wget.download('http://ufldl.stanford.edu/housenumbers/train.tar.gz', root_dir, bar=bar_custom)
        # extract
        print("\nExtracting files ...")
        with tarfile.open(train_compressed_path) as tar:
            tar.extractall(root_dir)

    if not os.path.exists(test_folder_path):
        # download
        if not os.path.exists(test_compressed_path):
            print("Downloading eval data to {}".format(root_dir))
            wget.download('http://ufldl.stanford.edu/housenumbers/test.tar.gz', root_dir, bar=bar_custom)
        # extract
        print("\nExtracting files ...")
        with tarfile.open(test_compressed_path) as tar:
            tar.extractall(root_dir)

    # glob and generate bbox files
    if not os.path.exists(train_file_path):
        print("\nConstructing bounding box data ...")
        _extract_metadata(train_folder_path, "train", train_file_path)
    if not os.path.exists(test_file_path):
        print("\nConstructing bounding box data ...")
        _extract_metadata(test_folder_path, "test", test_file_path)

    return PickleDataset(train_file_path), PickleDataset(test_file_path)