flowers.py 7.0 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
This module will download dataset from
16
http://www.robots.ox.ac.uk/~vgg/data/flowers/102/index.html
17 18
and parse train/test set intopaddle reader creators.

19
This set contains images of flowers belonging to 102 different categories.
20 21 22 23 24 25
The images were acquired by searching the web and taking pictures. There are a
minimum of 40 images for each category.

The database was used in:

Nilsback, M-E. and Zisserman, A. Automated flower classification over a large
26 27
 number of classes.Proceedings of the Indian Conference on Computer Vision,
Graphics and Image Processing (2008)
28 29 30 31
http://www.robots.ox.ac.uk/~vgg/publications/papers/nilsback08.{pdf,ps.gz}.

"""
import itertools
32
import functools
33 34
from common import download
import tarfile
J
Jiabin Yang 已提交
35 36
import six
from six.moves import cPickle as pickle
37
import scipy.io as scio
38 39
from paddle.dataset.image import *
from paddle.reader import *
40 41
import os
import numpy as np
42
from multiprocessing import cpu_count
43 44
__all__ = ['train', 'test', 'valid']

J
Jiabin Yang 已提交
45 46 47 48
DATA_URL = 'http://paddlemodels.cdn.bcebos.com/flowers/102flowers.tgz'
LABEL_URL = 'http://paddlemodels.cdn.bcebos.com/flowers/imagelabels.mat'
SETID_URL = 'http://paddlemodels.cdn.bcebos.com/flowers/setid.mat'
DATA_MD5 = '52808999861908f626f3c1f4e79d11fa'
49 50
LABEL_MD5 = 'e0620be6f572b9609742df49c70aed4d'
SETID_MD5 = 'a5357ecc9cb78c4bef273ce3793fc85c'
W
wanghaoshuang 已提交
51 52 53 54 55 56
# In official 'readme', tstid is the flag of test data
# and trnid is the flag of train data. But test data is more than train data.
# So we exchange the train data and test data.
TRAIN_FLAG = 'tstid'
TEST_FLAG = 'trnid'
VALID_FLAG = 'valid'
57 58


59
def default_mapper(is_train, sample):
60 61 62 63
    '''
    map image bytes data to type needed by model input layer
    '''
    img, label = sample
64
    img = load_image_bytes(img)
D
dangqingqing 已提交
65
    img = simple_transform(
D
dangqingqing 已提交
66
        img, 256, 224, is_train, mean=[103.94, 116.78, 123.68])
67 68 69
    return img.flatten().astype('float32'), label


70 71 72 73
train_mapper = functools.partial(default_mapper, True)
test_mapper = functools.partial(default_mapper, False)


74 75 76
def reader_creator(data_file,
                   label_file,
                   setid_file,
77
                   dataset_name,
78
                   mapper,
79
                   buffered_size=1024,
W
wanghaoshuang 已提交
80
                   use_xmap=True):
81
    '''
82
    1. read images from tar file and
83 84
        merge images into batch files in 102flowers.tgz_batch/
    2. get a reader to read sample from batch file
85 86

    :param data_file: downloaded data file
87
    :type data_file: string
88
    :param label_file: downloaded label file
89 90 91 92
    :type label_file: string
    :param setid_file: downloaded setid file containing information
                        about how to split dataset
    :type setid_file: string
93 94
    :param dataset_name: data set name (tstid|trnid|valid)
    :type dataset_name: string
95
    :param mapper: a function to map image bytes data to type
96 97
                    needed by model input layer
    :type mapper: callable
98 99
    :param buffered_size: the size of buffer used to process images
    :type buffered_size: int
100 101 102
    :return: data reader
    :rtype: callable
    '''
103 104 105 106 107 108 109
    labels = scio.loadmat(label_file)['labels'][0]
    indexes = scio.loadmat(setid_file)[dataset_name][0]
    img2label = {}
    for i in indexes:
        img = "jpg/image_%05d.jpg" % i
        img2label[img] = labels[i - 1]
    file_list = batch_images_from_tar(data_file, dataset_name, img2label)
110 111 112 113 114

    def reader():
        for file in open(file_list):
            file = file.strip()
            batch = None
J
Jiabin Yang 已提交
115 116 117 118 119
            with open(file, 'rb') as f:
                if six.PY2:
                    batch = pickle.load(f)
                else:
                    batch = pickle.load(f, encoding='bytes')
120 121 122
            data = batch['data']
            labels = batch['label']
            for sample, label in itertools.izip(data, batch['label']):
L
livc 已提交
123
                yield sample, int(label) - 1
124

W
wanghaoshuang 已提交
125
    if use_xmap:
C
chengduoZH 已提交
126 127
        cpu_num = int(os.environ.get('CPU_NUM', cpu_count()))
        return xmap_readers(mapper, reader, cpu_num, buffered_size)
128 129
    else:
        return map_readers(mapper, reader)
130 131


132
def train(mapper=train_mapper, buffered_size=1024, use_xmap=True):
133
    '''
134 135 136
    Create flowers training set reader.
    It returns a reader, each sample in the reader is
    image pixels in [0, 1] and label in [1, 102]
137 138 139 140 141 142
    translated from original color image by steps:
    1. resize to 256*256
    2. random crop to 224*224
    3. flatten
    :param mapper:  a function to map sample.
    :type mapper: callable
143 144
    :param buffered_size: the size of buffer used to process images
    :type buffered_size: int
145 146 147 148 149 150
    :return: train data reader
    :rtype: callable
    '''
    return reader_creator(
        download(DATA_URL, 'flowers', DATA_MD5),
        download(LABEL_URL, 'flowers', LABEL_MD5),
W
wanghaoshuang 已提交
151 152
        download(SETID_URL, 'flowers', SETID_MD5), TRAIN_FLAG, mapper,
        buffered_size, use_xmap)
153 154


155
def test(mapper=test_mapper, buffered_size=1024, use_xmap=True):
156
    '''
157 158 159
    Create flowers test set reader.
    It returns a reader, each sample in the reader is
    image pixels in [0, 1] and label in [1, 102]
160 161 162 163 164 165
    translated from original color image by steps:
    1. resize to 256*256
    2. random crop to 224*224
    3. flatten
    :param mapper:  a function to map sample.
    :type mapper: callable
166 167
    :param buffered_size: the size of buffer used to process images
    :type buffered_size: int
168 169 170 171 172 173
    :return: test data reader
    :rtype: callable
    '''
    return reader_creator(
        download(DATA_URL, 'flowers', DATA_MD5),
        download(LABEL_URL, 'flowers', LABEL_MD5),
W
wanghaoshuang 已提交
174 175
        download(SETID_URL, 'flowers', SETID_MD5), TEST_FLAG, mapper,
        buffered_size, use_xmap)
176 177


178
def valid(mapper=test_mapper, buffered_size=1024, use_xmap=True):
179
    '''
180 181 182
    Create flowers validation set reader.
    It returns a reader, each sample in the reader is
    image pixels in [0, 1] and label in [1, 102]
183 184 185 186
    translated from original color image by steps:
    1. resize to 256*256
    2. random crop to 224*224
    3. flatten
187 188 189 190 191 192
    :param mapper:  a function to map sample.
    :type mapper: callable
    :param buffered_size: the size of buffer used to process images
    :type buffered_size: int
    :return: test data reader
    :rtype: callable
193 194 195 196
    '''
    return reader_creator(
        download(DATA_URL, 'flowers', DATA_MD5),
        download(LABEL_URL, 'flowers', LABEL_MD5),
W
wanghaoshuang 已提交
197 198
        download(SETID_URL, 'flowers', SETID_MD5), VALID_FLAG, mapper,
        buffered_size, use_xmap)
199 200 201 202 203 204


def fetch():
    download(DATA_URL, 'flowers', DATA_MD5)
    download(LABEL_URL, 'flowers', LABEL_MD5)
    download(SETID_URL, 'flowers', SETID_MD5)