使用Mxnet训练图片分类类器
１、准备数据：
(1)建立一个根目录然后，再为每一个类别的图片建立一个子文件夹，把每一类图片放入对应的子文件夹即可。--root:
----class1
----class2
......
----classn
首先生成训练集和测试集的list,命令如下：
Python ~/mxnet/tools/im2rec.py －－list True －－recursive True －－train-ratio 0.9 myData /home/xxx/root/
–list：当要生成list文件时，这个参数一定要设为True，表示当前用来生成的list文件；默认是生成rec文件；
–recursive：递归的遍历你的所有数据集，要设为True；
–train-ratio：用来将你的全部数据集拆分成两部分：训练集（train）和交叉验证集（val），具体多少作为训练集，多少作为验证集，就由这个参数来确定；
–test-ratio：同上，分成训练集和测试集两部分；
–exts：这个是你数据的后缀（注，这里我们一般说的图片数据），目前的MXNet只支持两种图片格式：jpg和jpeg。
执行完这个命令，你会发现生成两个文件：myData_train.lst和myData_val.lst
(2)生成rec文件:
python ~/mxnet/tools/im2rec.py –num-thread 4 –pass-through 1 myData /home/xxx/root/
–pass-through: 设为1，即跳过矩阵变换，否则会报错：未知的array type；
myData就是第一步中生成.lst文件的前缀，这里用它来生成rec；
执行完这条命令，你就会看到两个文件：myData_train.rec和myData_val.rec
２、加载训练数据以及验证数据集：
def get_iterators(batch_size, data_shape=(3, 224, 224)):
train = mx.io.ImageRecordIter(
path_imgrec = './ld_train/my_images_train.rec',
data_name = 'data',
label_name = 'softmax_label',
batch_size = batch_size,
data_shape = data_shape,
shuffle = True,
rand_crop = True,
rand_mirror = True)
val = mx.io.ImageRecordIter(
path_imgrec = './ld_train/my_images_val.rec',
data_name = 'data',
label_name = 'softmax_label',
batch_size = batch_size,
data_shape = data_shape,
rand_crop = False,
rand_mirror = False)
return (train, val)
train,val=get_iterators(128,(3,128,128))＃指定Batch_size以及图片大小。３、定义网络结构：
这里以resnet为例：
'''
Adapted from https://github.com/tornadomeet/ResNet/blob/master/symbol_resnet.py
Original author Wei Wu
Implemented the following paper:
Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun. "Identity Mappings in Deep Residual Networks"
'''
import mxnet as mx
import numpy as np
def residual_unit(data, num_filter, stride, dim_match, name, bottle_neck=True, bn_mom=0.9, workspace=256, memonger=False):
"""Return ResNet Unit symbol for building ResNet
Parameters
----------
data : str
Input data
num_filter : int
Number of output channels
bnf : int
Bottle neck channels factor with regard to num_filter
stride : tupe
Stride used in convolution
dim_match : Boolen
True means channel number between input and output is the same, otherwise means differ
name : str
Base name of the operators
workspace : int
Workspace used in convolution operator
"""
if bottle_neck:
# the same as https://github.com/facebook/fb.resnet.torch#notes, a bit difference with origin paper
bn1 = mx.sym.BatchNorm(data=data, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn1')
act1 = mx.sym.Activation(data=bn1, act_type='relu', name=name + '_relu1')
weight = mx.symbol.Variable(name=name + '_conv1_weight', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
conv1 = mx.sym.Convolution(data=act1, weight=weight, num_filter=int(num_filter*0.25), kernel=(1,1), stride=(1,1), pad=(0,0),
no_bias=True, workspace=workspace, name=name + '_conv1')
bn2 = mx.sym.BatchNorm(data=conv1, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn2')
act2 = mx.sym.Activation(data=bn2, act_type='relu', name=name + '_relu2')
weight = mx.symbol.Variable(name=name + '_conv2_weight', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
conv2 = mx.sym.Convolution(data=act2, weight=weight, num_filter=int(num_filter*0.25), kernel=(3,3), stride=stride, pad=(1,1),
no_bias=True, workspace=workspace, name=name + '_conv2')
bn3 = mx.sym.BatchNorm(data=conv2, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn3')
act3 = mx.sym.Activation(data=bn3, act_type='relu', name=name + '_relu3')
weight = mx.symbol.Variable(name=name + '_conv3_weight', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
conv3 = mx.sym.Convolution(data=act3, weight=weight, num_filter=num_filter, kernel=(1,1), stride=(1,1), pad=(0,0), no_bias=True,
workspace=workspace, name=name + '_conv3')
if dim_match:
shortcut = data
else:
weight = mx.symbol.Variable(name=name + '_sc_weight', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
shortcut = mx.sym.Convolution(data=act1, weight=weight, num_filter=num_filter, kernel=(1,1), stride=stride, no_bias=True,
workspace=workspace, name=name+'_sc')
if memonger:
shortcut._set_attr(mirror_stage='True')
return conv3 + shortcut
else:
bn1 = mx.sym.BatchNorm(data=data, fix_gamma=False, momentum=bn_mom, eps=2e-5, name=name + '_bn1')
act1 = mx.sym.Activation(data=bn1, act_type='relu', name=name + '_relu1')
weight = mx.symbol.Variable(name=name + '_conv1_weight', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
conv1 = mx.sym.Convolution(data=act1, weight=weight, num_filter=num_filter, kernel=(3,3), stride=stride, pad=(1,1),
no_bias=True, workspace=workspace, name=name + '_conv1')
bn2 = mx.sym.BatchNorm(data=conv1, fix_gamma=False, momentum=bn_mom, eps=2e-5, name=name + '_bn2')
act2 = mx.sym.Activation(data=bn2, act_type='relu', name=name + '_relu2')
weight = mx.symbol.Variable(name=name + '_conv2_weight', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
conv2 = mx.sym.Convolution(data=act2, weight=weight, num_filter=num_filter, kernel=(3,3), stride=(1,1), pad=(1,1),
no_bias=True, workspace=workspace, name=name + '_conv2')
if dim_match:
shortcut = data
else:
weight = mx.symbol.Variable(name=name + '_sc_weight', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
shortcut = mx.sym.Convolution(data=act1, weight=weight, num_filter=num_filter, kernel=(1,1), stride=stride, no_bias=True,
workspace=workspace, name=name+'_sc')
if memonger:
shortcut._set_attr(mirror_stage='True')
return conv2 + shortcut
def resnet(units, num_stages, filter_list, num_classes, image_shape, bottle_neck=True, bn_mom=0.9, workspace=256, memonger=False):
"""Return ResNet symbol of
Parameters
----------
units : list
Number of units in each stage
num_stages : int
Number of stage
filter_list : list
Channel size of each stage
num_classes : int
Ouput size of symbol
dataset : str
Dataset type, only cifar10 and imagenet supports
workspace : int
Workspace used in convolution operator
"""
num_unit = len(units)
assert(num_unit == num_stages)
data = mx.sym.Variable(name='data')
data = mx.symbol.Cast(data=data, dtype=np.float16)
data = mx.sym.BatchNorm(data=data, fix_gamma=True, eps=2e-5, momentum=bn_mom, name='bn_data')
(nchannel, height, width) = image_shape
weight = mx.symbol.Variable(name='conv0_weight', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
if height <= 32: # such as cifar10
body = mx.sym.Convolution(data=data, weight=weight, num_filter=filter_list[0], kernel=(3, 3), stride=(1,1), pad=(1, 1),
no_bias=True, name="conv0", workspace=workspace)
else: # often expected to be 224 such as imagenet
body = mx.sym.Convolution(data=data, weight=weight, num_filter=filter_list[0], kernel=(7, 7), stride=(2,2), pad=(3, 3),
no_bias=True, name="conv0", workspace=workspace)
body = mx.sym.BatchNorm(data=body, fix_gamma=False, eps=2e-5, momentum=bn_mom, name='bn0')
body = mx.sym.Activation(data=body, act_type='relu', name='relu0')
body = mx.symbol.Pooling(data=body, kernel=(3, 3), stride=(2,2), pad=(1,1), pool_type='max')
for i in range(num_stages):
body = residual_unit(body, filter_list[i+1], (1 if i==0 else 2, 1 if i==0 else 2), False,
name='stage%d_unit%d' % (i + 1, 1), bottle_neck=bottle_neck, workspace=workspace,
memonger=memonger)
for j in range(units[i]-1):
body = residual_unit(body, filter_list[i+1], (1,1), True, name='stage%d_unit%d' % (i + 1, j + 2),
bottle_neck=bottle_neck, workspace=workspace, memonger=memonger)
bn1 = mx.sym.BatchNorm(data=body, fix_gamma=False, eps=2e-5, momentum=bn_mom, name='bn1')
relu1 = mx.sym.Activation(data=bn1, act_type='relu', name='relu1')
# Although kernel is not used here when global_pool=True, we should put one
pool1 = mx.symbol.Pooling(data=relu1, global_pool=True, kernel=(7, 7), pool_type='avg', name='pool1')
flat = mx.symbol.Flatten(data=pool1)
weight = mx.symbol.Variable(name='fc1_weight', dtype=np.float32)
bias = mx.symbol.Variable(name='fc1_bias', dtype=np.float32)
weight = mx.symbol.Cast(data=weight, dtype=np.float16)
bias = mx.symbol.Cast(data=bias, dtype=np.float16)
fc1 = mx.symbol.FullyConnected(data=flat, weight=weight, bias=bias, num_hidden=num_classes, name='fc1')
fc1 = mx.symbol.Cast(data=fc1, dtype=np.float32)
return mx.symbol.SoftmaxOutput(data=fc1, name='softmax')
def get_symbol(num_classes, num_layers, image_shape, conv_workspace=256, **kwargs):
"""
Adapted from https://github.com/tornadomeet/ResNet/blob/master/train_resnet.py
Original author Wei Wu
"""
image_shape = [int(l) for l in image_shape.split(',')]
(nchannel, height, width) = image_shape
if height <= 28:
num_stages = 3
if (num_layers-2) % 9 == 0 and num_layers >= 164:
per_unit = [(num_layers-2)//9]
filter_list = [16, 64, 128, 256]
bottle_neck = True
elif (num_layers-2) % 6 == 0 and num_layers < 164:
per_unit = [(num_layers-2)//6]
filter_list = [16, 16, 32, 64]
bottle_neck = False
else:
raise ValueError("no experiments done on num_layers {}, you can do it youself".format(num_layers))
units = per_unit * num_stages
else:
if num_layers >= 50:
filter_list = [64, 256, 512, 1024, 2048]
bottle_neck = True
else:
filter_list = [64, 64, 128, 256, 512]
bottle_neck = False
num_stages = 4
if num_layers == 18:
units = [2, 2, 2, 2]
elif num_layers == 34:
units = [3, 4, 6, 3]
elif num_layers == 50:
units = [3, 4, 6, 3]
elif num_layers == 101:
units = [3, 4, 23, 3]
elif num_layers == 152:
units = [3, 8, 36, 3]
elif num_layers == 200:
units = [3, 24, 36, 3]
elif num_layers == 269:
units = [3, 30, 48, 8]
else:
raise ValueError("no experiments done on num_layers {}, you can do it youself".format(num_layers))
return resnet(units = units,
num_stages = num_stages,
filter_list = filter_list,
num_classes = num_classes,
image_shape = image_shape,
bottle_neck = bottle_neck,
workspace = conv_workspace)
resnet18=get_symbol(2,18,'3,128,128')４、训练并保存模型：
# construct a callback function to save checkpoints
model_prefix = 'resnet18'
checkpoint = mx.callback.do_checkpoint(model_prefix)
mod = mx.mod.Module(symbol=lenet,context=mx.gpu(0))
mod.fit(train,
eval_data=val,
optimizer='sgd',
optimizer_params={'learning_rate':0.001, 'momentum': 0.9},
eval_metric='acc',
num_epoch=30,
epoch_end_callback=checkpoint)５、加载模型准备测试：
sym, arg_params, aux_params = mx.model.load_checkpoint('resnet18', 30)
mod = mx.mod.Module(symbol=sym, context=mx.gpu(0), label_names=None)
mod.bind(for_training=False, data_shapes=[('data', (1,3,128,128))],
label_shapes=mod._label_shapes)
mod.set_params(arg_params, aux_params, allow_missing=True)６、测试数进行预测：
%matplotlib inline
import matplotlib.pyplot as plt
import cv2
import numpy as np
# define a simple data batch
from collections import namedtuple
Batch = namedtuple('Batch', ['data'])
def get_image(path, show=False):
# download and show the image
img = cv2.cvtColor(cv2.imread(path), cv2.COLOR_BGR2RGB)
if img is None:
return None
if show:
plt.imshow(img)
plt.axis('off')
# convert into format (batch, RGB, width, height)
img = cv2.resize(img, (128, 128))
img = np.swapaxes(img, 0, 2)
img = np.swapaxes(img, 1, 2)
img = img[np.newaxis, :]
return img
def predict(path):
img = get_image(path, show=True)
# compute the predict probabilities
mod.forward(Batch([mx.nd.array(img)]))
prob = mod.get_outputs()[0].asnumpy()
# print the top-5
prob = np.squeeze(prob)
a = np.argsort(prob)[::-1]
for i in a[0:5]:
print((prob[i]))
predict('/home/test_images/cat.jpg')