image-augmentation
April 9, 2019
1 Image Augmentation
In [1]: %matplotlib inline
import d2l
import mxnet as mx
from mxnet import autograd, gluon, image, init, nd
from mxnet.gluon import data as gdata, loss as gloss, utils as gutils
import sys
import time
1.1 Setup
Load a sample image with a shape of 400 × 500.
In [2]: d2l.set_figsize()
img = image.imread('cat1.jpg')
d2l.plt.imshow(img.asnumpy())
Out[2]: <matplotlib.image.AxesImage at 0x7fa3945cd2d0>
1
1.1.1 The Drawing Function show_images
In [3]: def show_images(imgs, num_rows, num_cols, scale=2):
figsize = (num_cols * scale, num_rows * scale)
_, axes = d2l.plt.subplots(num_rows, num_cols, figsize=figsize)
for i in range(num_rows):
for j in range(num_cols):
axes[i][j].imshow(imgs[i * num_cols + j].asnumpy())
axes[i][j].axes.get_xaxis().set_visible(False)
axes[i][j].axes.get_yaxis().set_visible(False)
return axes
1.1.2 Apply an Augmentation Multiple Times and Draw Results
In [4]: def apply(img, aug, num_rows=2, num_cols=4, scale=2):
Y = [aug(img) for _ in range(num_rows * num_cols)]
show_images(Y, num_rows, num_cols, scale)
1.2 Flip and Crop
1.2.1 Randomly Flip Left and Right
In [5]: apply(img, gdata.vision.transforms.RandomFlipLeftRight())
1.2.2 Randomly Flip Top and Bottom
In [6]: apply(img, gdata.vision.transforms.RandomFlipTopBottom())
2
1.2.3 Randomly Crop a Region
In [7]: shape_aug = gdata.vision.transforms.RandomResizedCrop(
(200, 200), scale=(0.1, 1), ratio=(0.5, 2))
apply(img, shape_aug)
1.3 Color
1.3.1 Randomly Change Brightness
In [8]: apply(img, gdata.vision.transforms.RandomBrightness(0.5))
3
1.3.2 Randomly Change Hue
In [9]: apply(img, gdata.vision.transforms.RandomHue(0.5))
1.3.3 Randomly Change Brightness, Contrast, Saturation and Hue
In [10]: color_aug = gdata.vision.transforms.RandomColorJitter(
brightness=0.5, contrast=0.5, saturation=0.5, hue=0.5)
apply(img, color_aug)
4
1.4 Use Multiple Image Augmentation Methods Together
In [21]: augs = gdata.vision.transforms.Compose([
gdata.vision.transforms.RandomFlipLeftRight(),
gdata.vision.transforms.RandomColorJitter(
brightness=0.5, contrast=0.5, saturation=0.5, hue=0.5),
gdata.vision.transforms.RandomResizedCrop(
(200, 200), scale=(0.1, 1), ratio=(0.5, 2))])
apply(img, augs)
5
1.5 Using Image Augmentations to Train Models
In [12]: show_images(gdata.vision.CIFAR10(train=True)[0:32][0], 4, 8, scale=0.8)
train_augs = gdata.vision.transforms.Compose([
gdata.vision.transforms.RandomFlipLeftRight(),
gdata.vision.transforms.ToTensor()])
test_augs = gdata.vision.transforms.Compose([
gdata.vision.transforms.ToTensor()])
1.5.1 Load Data
In [13]: num_workers = 0 if sys.platform.startswith('win32') else 4
def load_cifar10(is_train, augs, batch_size):
return gdata.DataLoader(
gdata.vision.CIFAR10(train=is_train).transform_first(augs),
batch_size=batch_size, shuffle=is_train, num_workers=num_workers)
1.5.2 Use Multi-GPUs
In [14]: def try_all_gpus():
ctxes = []
try:
for i in range(16):
ctx = mx.gpu(i)
_ = nd.array([0], ctx=ctx)
ctxes.append(ctx)
except mx.base.MXNetError:
pass
6
if not ctxes:
ctxes = [mx.cpu()]
return ctxes
1.5.3 Split Data into Multi-GPUs
In [15]: def _get_batch(batch, ctx):
features, labels = batch
if labels.dtype != features.dtype:
labels = labels.astype(features.dtype)
return (gutils.split_and_load(features, ctx),
gutils.split_and_load(labels, ctx), features.shape[0])
1.5.4 Evaluate on Multi-GPUs
In [16]: def evaluate_accuracy(data_iter, net, ctx=[mx.cpu()]):
if isinstance(ctx, mx.Context):
ctx = [ctx]
acc_sum, n = nd.array([0]), 0
for batch in data_iter:
features, labels, _ = _get_batch(batch, ctx)
for X, y in zip(features, labels):
y = y.astype('float32')
acc_sum += (net(X).argmax(axis=1) == y).sum().copyto(mx.cpu())
n += y.size
acc_sum.wait_to_read()
return acc_sum.asscalar() / n
1.5.5 Train on Multi-GPUs
In [17]: def train(train_iter, test_iter, net, loss, trainer, ctx, num_epochs):
print('training on', ctx)
if isinstance(ctx, mx.Context):
ctx = [ctx]
for epoch in range(num_epochs):
train_l_sum, train_acc_sum, n, m, start = 0.0, 0.0, 0, 0, time.time()
for i, batch in enumerate(train_iter):
Xs, ys, batch_size = _get_batch(batch, ctx)
ls = []
with autograd.record():
y_hats = [net(X) for X in Xs]
ls = [loss(y_hat, y) for y_hat, y in zip(y_hats, ys)]
for l in ls:
l.backward()
trainer.step(batch_size)
train_l_sum += sum([l.sum().asscalar() for l in ls])
n += sum([l.size for l in ls])
train_acc_sum += sum([(y_hat.argmax(axis=1) == y).sum().asscalar()
7
for y_hat, y in zip(y_hats, ys)])
m += sum([y.size for y in ys])
test_acc = evaluate_accuracy(test_iter, net, ctx)
print('epoch %d, loss %.4f, train acc %.3f, test acc %.3f, '
'time %.1f sec'
% (epoch + 1, train_l_sum / n, train_acc_sum / m, test_acc,
time.time() - start))
1.5.6 A Function to Train with Various Augmentations
In [18]: def train_with_data_aug(train_augs, test_augs, lr=0.001):
batch_size, ctx, net = 256, try_all_gpus(), d2l.resnet18(10)
net.initialize(ctx=ctx, init=init.Xavier())
trainer = gluon.Trainer(net.collect_params(), 'adam',
{'learning_rate': lr})
loss = gloss.SoftmaxCrossEntropyLoss()
train_iter = load_cifar10(True, train_augs, batch_size)
test_iter = load_cifar10(False, test_augs, batch_size)
train(train_iter, test_iter, net, loss, trainer, ctx, num_epochs=8)
1.5.7 Train with Image Augmentation
In [19]: train_with_data_aug(train_augs, test_augs)
('training on', [gpu(0), gpu(1)])
epoch 1, loss 1.4165, train acc 0.498, test acc 0.612, time 39.1 sec
epoch 2, loss 0.8359, train acc 0.702, test acc 0.705, time 35.3 sec
epoch 3, loss 0.6084, train acc 0.787, test acc 0.775, time 35.4 sec
epoch 4, loss 0.4859, train acc 0.831, test acc 0.815, time 35.2 sec
epoch 5, loss 0.3986, train acc 0.862, test acc 0.765, time 35.2 sec
epoch 6, loss 0.3328, train acc 0.885, test acc 0.817, time 35.1 sec
epoch 7, loss 0.2772, train acc 0.904, test acc 0.838, time 35.2 sec
epoch 8, loss 0.2343, train acc 0.919, test acc 0.833, time 35.1 sec
In [20]: train_with_data_aug(test_augs, test_augs)
('training on', [gpu(0), gpu(1)])
epoch 1, loss 1.3561, train acc 0.518, test acc 0.527, time 35.5 sec
epoch 2, loss 0.7933, train acc 0.720, test acc 0.692, time 35.3 sec
epoch 3, loss 0.5731, train acc 0.799, test acc 0.751, time 35.3 sec
epoch 4, loss 0.4256, train acc 0.850, test acc 0.762, time 35.3 sec
epoch 5, loss 0.3102, train acc 0.892, test acc 0.732, time 35.2 sec
epoch 6, loss 0.2315, train acc 0.919, test acc 0.793, time 35.1 sec
epoch 7, loss 0.1623, train acc 0.943, test acc 0.734, time 35.1 sec
epoch 8, loss 0.1142, train acc 0.960, test acc 0.800, time 35.1 sec
