import mxnet as mx
import numpy as np
from dataset import get_data
from symbols import symbols
import pandas as pd
batch_size = 1
num_test_images = 10 #
# get data csv
mnistcsv = get_data.read(one_hot=True)
# get model
net = symbols.get_model('simple2', pretrained=True)
#net.load_parameters( os.path.join('symbols','para','simple0.params') )
for i in range(num_test_images):
X, _ = mnistcsv.validation.next_batch(batch_size)
X = nd.array(X)
y = net(X)
ans = y.argmax(axis=1).asnumpy()
print("%d-th type %d" % (i, ans))
# ===================== ====================
img_data = nd.array(mnistcsv.validation.data)
y = net(img_data)
Label = y.argmax(axis=1).asnumpy().astype(np.int8)
Image = np.arange(len(Label)) + 1
from symbols import symbols
import pandas as pd
from gluoncv import model_zoo as mzoo
batch_size = 1
num_test_images = 10 #
ctx = mx.cpu(
) #[mx.gpu(i) for i in range(num_gpus)] if num_gpus > 0 else [mx.cpu()]
# get data csv
mnistcsv = get_data.read(one_hot=True)
# get model
modelname = 'mobilenet0.25'
net = symbols.get_model('mobilenet0.25', pretrained=True)
for i in range(num_test_images):
X, _ = mnistcsv.validation.next_batch(batch_size)
X = nd.array(X)
X = X.reshape((-1, 1, 28, 28))
X = nd.concat(X, X, X, dim=1)
y = net(X)
ans = y.argmax(axis=1).asnumpy()
print("%d-th type %d" % (i, ans))
'''
0-th type 2
1-th type 0
2-th type 9
3-th type 0
4-th type 3
transform = lambda data, label: (data.reshape(784, ).astype(np.float32) / 255,
label)
train_data = gluon.data.DataLoader(dataset=gluon.data.vision.MNIST(
train=True, transform=transform),
batch_size=100,
shuffle=True,
last_batch='discard')
val_data = gluon.data.DataLoader(dataset=gluon.data.vision.MNIST(
train=False, transform=transform),
batch_size=100,
shuffle=False)
# network
modelname = 'semi_pi_simple2'
basemodel_zoo = 'simple2'
net = symbols.get_model('simple2')
net.initialize(mx.init.Xavier(magnitude=2.24))
#net.load_parameters(os.path.join('symbols','para','%s.params'%(modelname)))
# g(x) : stochastic input augmentation function
def g(x):
return x + nd.random.normal(0, stochastic_ratio, shape=x.shape)
# loss function
l_logistic = gloss.SoftmaxCrossEntropyLoss()
l_l2loss = gloss.L2Loss()
metric = mx.metric.Accuracy()