Get from chainer's dataset module, the resulting images is 0-255 uint8 numpy matrix, saved into pyarrow zero-copy data type

usage:

from dataset.minist.feed import feed
path = "./save/"
feed(feed_path=path)

then training data will be stored in ./save/X.pa, the training label will be stored in ./save/Y.pa

to load the data into numpy, try:

from dataset import pa2np
X, Y = pa2np("./save/X.pa"), pa2np("./save/Y.pa")

sample of resulting dataset:

Get from chainer's dataset moduel, the resulting image is 0-255 uint8 numpy matrix, saved into pyarrow zero-copy data type

usage:

for cifar-10

from dataset.cifar.feed import feed
path = "./save/"
feed(feed_path=path, dataset_type=10)

then training data will be stored in ./save/X_10.pa, the training label will be stored in ./save/Y_10.pa

sample of resulting dataset:

for cifar-100

from dataset.cifar.feed import feed
path = "./save/"
feed(feed_path=path, dataset_type=100)

then training data will be stored in ./save/X_100.pa, the training label will be stored in ./save/Y_100.pa

sample of resulting dataset:

To load the data, see that in mnist above

the resulting image is 0-255 uint8 numpy matrix, saved into pyarrow zero-copy data type

usage:

for unprocessed

from dataset.coil20.feed import feed
path = "./save/"
feed(feed_path=path, dataset_type='unprocessed')

then training data will be stored in ./save/X_unprocessed.pa, the training label will be stored in ./save/Y_unprocessed.pa

sample of resulting dataset:

for processed

from dataset.coil20.feed import feed
path = "./save/"
feed(feed_path=path, dataset_type='processed')

then training data will be stored in ./save/X_processed.pa, the training label will be stored in ./save/Y_processed.pa

To load the data, see that in mnist above

sample of resulting dataset:

below are some kaggle dataset

the resulting image is 0-255 uint8 numpy matrix, saved into pyarrow zero-copy data type

usage:

from dataset.kaggle.fer2013.feed import feed
path = "./save/"
feed(feed_path=path)

then training data will be stored in ./save/X.pa, the training label will be stored in ./save/Y.pa

sample of resulting dataset:

the resulting image and correponding label is save into dict object

usage:

from dataset.kaggle.whales2018.feed import feed
path = "./save/"
feed(feed_path=path)

usage:

from dataset.isbi.c2012.feed import feed, load 
path="./save/"
feed(feed_path=path)
#to get data:
data, mask = load(load_path=path)