def proc_and_dump_sup_test_data(sub_set_data, split, sup_size, used_labels):
all_images = sub_set_data["images"]
all_labels = sub_set_data["labels"]
# Added by Fuwen
images, labels = resampling_sup_by_labels(all_images, all_labels,
used_labels)
chosen_images = images
chosen_labels = labels
# print('test label range', np.amax(chosen_labels), np.amin(chosen_labels))
chosen_images = chosen_images / 255.0
mean, std = augmentation_transforms.get_mean_and_std()
chosen_images = (chosen_images - mean) / std
example_list = []
for image, label in zip(chosen_images, chosen_labels):
# Write example to the tfrecord file
example = tf.train.Example(features=tf.train.Features(
feature={
"image": _float_feature(image.reshape(-1)),
"label": _int64_feature(label.reshape(-1))
}))
example_list += [example]
out_path = os.path.join(FLAGS.output_base_dir,
format_sup_filename(split, sup_size))
tf.logging.info(">> saving {} {} examples to {}".format(
len(example_list), split, out_path))
save_tfrecord(example_list, out_path)
def proc_and_dump_sup_data(sub_set_data, split, sup_size=-1):
images = sub_set_data["images"]
labels = sub_set_data["labels"]
if sup_size != -1:
chosen_images, chosen_labels = get_data_by_size_lim(
images, labels, sup_size)
else:
chosen_images = images
chosen_labels = labels
chosen_images = chosen_images / 255.0
mean, std = augmentation_transforms.get_mean_and_std()
chosen_images = (chosen_images - mean) / std
example_list = []
for image, label in zip(chosen_images, chosen_labels):
# Write example to the tfrecord file
example = tf.train.Example(features=tf.train.Features(
feature={
"image": _float_feature(image.reshape(-1)),
"label": _int64_feature(label.reshape(-1))
}))
example_list += [example]
out_path = os.path.join(
FLAGS.output_base_dir,
format_sup_filename(split, sup_size)
)
tf.logging.info(">> saving {} {} examples to {}".format(
len(example_list), split, out_path))
save_tfrecord(example_list, out_path)
def save_processed_images_and_labels(chosen_images, chosen_labels, split):
chosen_images = chosen_images / 255.0
mean, std = augmentation_transforms.get_mean_and_std()
chosen_images = (chosen_images - mean) / std
example_list = []
for image, label in zip(chosen_images, chosen_labels):
# Write example to the tfrecord file
example = tf.train.Example(features=tf.train.Features(
feature={
"image": _float_feature(image.reshape(-1)),
"label": _int64_feature(label.reshape(-1))
}))
example_list += [example]
out_path = os.path.join(FLAGS.output_base_dir,
format_sup_filename(split, len(example_list)))
tf.logging.info(">> saving {} {} examples to {}".format(
len(example_list), split, out_path))
save_tfrecord(example_list, out_path)
def proc_and_dump_unsup_data(sub_set_data, aug_copy_num, used_labels):
ori_images = sub_set_data["images"].copy()
# print('used_labels', used_labels)
ori_images = resampling_unsup_by_labels(ori_images,
sub_set_data["labels"].copy(),
used_labels)
print('unsup ori_images', ori_images.shape)
image_idx = np.arange(len(ori_images))
np.random.shuffle(image_idx)
ori_images = ori_images[image_idx]
# tf.logging.info("first 5 indexes after shuffling: {}".format(
# str(image_idx[:5])))
ori_images = ori_images / 255.0
mean, std = augmentation_transforms.get_mean_and_std()
ori_images = (ori_images - mean) / std
if FLAGS.task_name == "cifar10":
aug_policies = found_policies.randaug_policies()
elif FLAGS.task_name == "svhn":
aug_policies = found_policies.randaug_policies()
example_list = []
for image in ori_images:
chosen_policy = aug_policies[np.random.choice(len(aug_policies))]
aug_image = augmentation_transforms.apply_policy(chosen_policy, image)
aug_image = augmentation_transforms.cutout_numpy(aug_image)
# Write example to the tfrecord file
example = tf.train.Example(features=tf.train.Features(
feature={
"ori_image": _float_feature(image.reshape(-1)),
"aug_image": _float_feature(aug_image.reshape(-1)),
}))
example_list += [example]
out_path = os.path.join(
FLAGS.output_base_dir,
format_unsup_filename(aug_copy_num),
)
save_tfrecord(example_list, out_path)