Caffe 训练时 accuracy precision始终为1是什么原因

> Caffe 实例笔记 1 CaffeNet从训练到分类及可视化参数特征 微调
Caffe 实例笔记 1 CaffeNet从训练到分类及可视化参数特征 微调
本文主要分四部分
1. 在命令行进行训练
2. 使用pycaffe进行分类及特征可视化
3. 进行微调,将caffenet使用在图片风格的预测上
1 使用caffeNet训练自己的数据集
主要参考:
官方网址:
数据集及第一部分参考网址:
主要步骤:
1. 准备数据集
2. 标记数据集
3. 创建lmdb格式的数据
4. 计算均值
5. 设置网络及求解器
6. 运行求解
由于imagenet的数据集太大,博主电脑显卡840m太弱,所以就选择了第二个网址中的数据集
,其训练集为1000张10类图片,验证集为200张图片,原作者已经整理好其标签放于对应的txt文件中,所以这里就省去上面的1-2步骤。
1.1 创建lmdb
使用对应的数据集创建lmdb:
这里使用 examples/imagenet/create_imagenet.sh,需要更改其路径和尺寸设置的选项,为了减小更改的数目,这里并没有自己新创建一个文件夹,而是直接使用了原来的imagenet的文件夹,而且将train.txt,val.txt都放置于/data/ilsvrc12中,
TRAIN_DATA_ROOT=/home/beatree/caffe-rc3/examples/imagenet/train/
VAL_DATA_ROOT=/home/beatree/caffe-rc3/examples/imagenet/val/
RESIZE=true
注意下面的地址的含义:
echo &Creating train lmdb...&
GLOG_logtostderr=1 $TOOLS/convert_imageset \
--resize_height=$RESIZE_HEIGHT \
--resize_width=$RESIZE_WIDTH \
--shuffle \
$TRAIN_DATA_ROOT \
$DATA/train.txt \
$EXAMPLE/ilsvrc12_train_lmdb
主要用了tools里的convert_imageset
1.2 计算均值
模型需要我们从每张图片减去均值,所以我们需要获得训练的均值,直接利用./examples/imagenet/make_imagenet_mean.sh创建均值文件binaryproto,如果之前创建了新的路径,这里同样需要修改sh文件里的路径。
这里的主要语句是
$TOOLS/compute_image_mean $EXAMPLE/ilsvrc12_train_lmdb \
$DATA/imagenet_mean.binaryproto
如果显示Check failed: size_in_datum == data_size () Incorrect data field size说明上一步的图片没有统一尺寸
1.3 设置网络及求解器
这里是利用原文的网络设置tain_val.prototxt和slover.prototext,在models/bvlc_reference_caffenet/solver.prototxt路径中,这里的训练和验证的网络基本一样用 include { phase: TRAIN } or include { phase: TEST }和来区分,其两点不同之处具体为:
transform_param { mirror: true#不同1:训练集会randomly mirrors the input image crop_size: 227 mean_file: &data/ilsvrc12/imagenet_mean.binaryproto& }
data_param { source: &examples/imagenet/ilsvrc12_train_lmdb&#不同2:来源不同 batch_size: 32#原文很大,显卡比较弱的会内存不足,这里改为了32,这里根据需要更改,验证集和训练集的设置也不一样 backend: LMDB }
另外在输出层也有不同,训练时的loss需要用来进行反向传递,而val就不需要了。
solver.protxt的改动:
net: &/home/beatree/caffe-rc3/examples/imagenet/train_val.prototxt&#网络配置存放地址
test_iter: 4, 每个批次是50,一共200个
test_interval: 300 #每300次测试一次
base_lr: 0.01 #是基础学习率,因为数据量小,0.01 就会下降太快了,因此可以改成 0.001,这里博主没有改
lr_policy: &step& #lr可以变化
gamma: 0.1 #学习率变化的比率
stepsize: 300
display: 20 #20层显示一次
max_iter: 1200 一共迭代1200次
momentum: 0.9
weight_decay: 0.0005
snapshot: 600 #每600存一个状态
snapshot_prefix: &/home/beatree/caffe-rc3/examples/imagenet/&#状态存放地址
使用上面的配置训练,得到的结果准确率仅仅是0.2+,数据集的制作者迭代了12000次得到0.5的准确率
1.5.1杀掉正在运行的caffe进程:
ps -A#查看所有进程,及caffe的代码
kill -9 代码#杀掉caffe
1.5.2 查看gpu的使用情况
nvidia-sim -l
(NVIDIA System Management Interface)
1.5.3 查看时间使用情况
./build/tools/caffe time --model=models/bvlc_reference_caffenet/train_val.prototxt
我的时间使用情况
Average Forward pass: 3490.86 ms.
Average Backward pass: 5666.73 ms.
Average Forward-Backward: 9157.66 ms.
Total Time: 457883 ms.
1.5.4 恢复数据
如果我们在训练途中就停电或者有了其他的情况,我们可以通过之前保存的状态恢复数据,使用的时候直接添加–snapshot参数即可,如:
./build/tools/caffe train --solver=models/bvlc_reference_caffenet/solver.prototxt --snapshot=models/bvlc_reference_caffenet/caffenet_train_iter_10000.solverstate
这时候运行会从snapshot开始继续运行,如从第600迭代时运行:
1.5.5 c++ 提取特征
when everything necessary is in place:
./build/tools/extract_features.bin models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel examples/_temp/imagenet_val.prototxt fc7 examples/_temp/features 10 leveldb
the features are stored to LevelDB examples/_temp/features.
1.5.6 使用c++分类
对于c++的学习应该读读tools/caffe.cpp里的代码。
其分类命令如下:
./build/examples/cpp_classification/classification.bin \ models/bvlc_reference_caffenet/deploy.prototxt \ models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel \ data/ilsvrc12/imagenet_mean.binaryproto \ data/ilsvrc12/synset_words.txt \ examples/images/cat.jpg
2 使用pycaffe分类
2.1 import
首先载入环境:
# set up Python environment: numpy for numerical routines, and matplotlib for plotting import numpy as np import matplotlib.pyplot as plt # display plots in this notebook %matplotlib inline#这里由于ipython启动时移除了 pylab 启动参数,所以需要使用这种格式查看,官网介绍http://ipython.org/ipython-doc/stable/interactive/reference.html#plotting-with-matplotlib: #To start IPython with matplotlib support, use the --matplotlib switch. If IPython is already running, you can run the %matplotlib magic. If no arguments are given, IPython will automatically detect your choice of matplotlib backend. You can also request a specific backend with %matplotlib backend, where backend must be one of: ‘tk’, ‘qt’, ‘wx’, ‘gtk’, ‘osx’. In the web notebook and Qt console, ‘inline’ is also a valid backend value, which produces static figures inlined inside the application window instead of matplotlib’s interactive figures that live in separate windows. # set display defaults #关于rcParams函数http://matplotlib.org/api/matplotlib_configuration_api.html#matplotlib.rcParams plt.rcParams['figure.figsize'] = (10, 10) # large images plt.rcParams['image.interpolation'] = 'nearest' # don't interpolate: show square pixels plt.rcParams['image.cmap'] = 'gray' # use grayscale output rather than a (potentially misleading) color heatmap
import caffe#如果没有设置好路径可能发现不了caffe,需要import sys cafe_root='你的路径',sys.path.insert(0,caffe_root+'python')之后再import caffe
下面下载模型,由于上面刚开始我们用的数据不是imagenet,现在我们直接下载一个模型,可能你的python中没有yaml,这里可以用pip安装(终端里):
sudo apt-get install python-pip
pip install pyyaml
cd #你的caffe root
./scripts/download_model_binary.py /home/beatree/caffe-rc3/model
/bvlc_reference_caffenet
#其他的网络路径如下:models/bvlc_alexnet models/bvlc_reference_rcnn_ilsvrc13 models/bvlc_googlenet model zoo的连接http://caffe.berkeleyvision.org/model_zoo.html,模型一共232m
2.2 模型载入
caffe.set_mode_cpu()#使用cpu模式
model_def='/home/beatree/caffe-rc3/models/bvlc_reference_caffenet/deploy.prototxt'
model_weights='/home/beatree/caffe-rc3/models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel'
net=caffe.Net(model_def,
model_weights,
caffe.TEST)
mu=np.load('/home/beatree/caffe-rc3/python/caffe/imagenet/ilsvrc_2012_mean.npy')
mu=mu.mean(1).mean(1)
mu长成下面这个样子:
array([[[ 110.,
110., ...,
110.9342804 ,
110.5134201 ],
111., ...,
110.6951828 ],
[ 110.525177
111., ...,
得到bgr的均值
print 'mean-subtracted values:', zip('BGR', mu)
mean-subtracted values: [('B', 104.9), ('G', 116.67), ('R', 122.6)]
matplotlib加载的image是像素[0-1],图片的数据格式[weight,high,channels],RGB 而caffe加载的图片需要的是[0-255]像素,数据格式[channels,weight,high],BGR,那么就需要转换 ,这里用了 caffe.io.Transformer,可以使用help()来获得相关信息,他的功能有
preprocess(self, in_, data)
set_channel_swap(self, in_, order)
set_input_scale(self, in_, scale)
set_mean(self, in_, mean)
set_raw_scale(self, in_, scale)
set_transpose(self, in_, order)
# create transformer for the input called 'data' transformer = caffe.io.Transformer({'data': net.blobs['data'].data.shape})#net.blobs['data'].data.shape=(10, 3, 227, 227) transformer.set_transpose('data', (2,0,1)) # move image channels to outermost dimension第一个变成了channels transformer.set_mean('data', mu) # subtract the dataset-mean value in each channel transformer.set_raw_scale('data', 255) # rescale from [0, 1] to [0, 255] transformer.set_channel_swap('data', (2,1,0)) # swap channels from RGB to BGR
2.3 cpu 分类
这里可以准备开始分类了,下面改变输入size的步骤也可以跳过,这里batchsize设置为50只是为了演示用,实际我们只对一张图片进行分类。
# set the size of the input (we can skip this if we're happy
# we can also change it later, e.g., for different batch sizes)
net.blobs['data'].reshape(50,
# batch size
# 3-channel (BGR) images
# image size is 227x227
image = caffe.io.load_image( 'path/to/images/cat.jpg')
transformed_image = transformer.preprocess('data', image)
plt.imshow(image)
得到一个可爱的小猫,接下来看一看模型是不是认为她是不是小猫
# copy the image data into the memory allocated for the net
net.blobs['data'].data[...] = transformed_image
### perform classification
output = net.forward()
output_prob = output['prob'][0]
# the output probability vector for the first image in the batch
print 'predicted class is:', output_prob.argmax(),output_prob[output_prob.argmax()]
得到结果:
predicted calss is 281 0.312436
也就是第281种最有可能,概率比重是0.312436
那么第231种是不是猫呢,让我们接着看
# load ImageNet labels
labels_file = caffe_root + 'data/ilsvrc12/synset_words.txt'#如果没有这个文件,须运行/data/ilsvrc12/get_ilsvrc_aux.sh
labels = np.loadtxt(labels_file, str, delimiter='\t')
print 'output label:', labels[output_prob.argmax()]
结果是answer is n tabby, tabby cat连花纹都判断对了。接下来让我们进一步观察判断的结果:
# sort top five predictions from softmax output
top_inds = output_prob.argsort()[::-1][:5]
# reverse sort and take five largest items
print 'probabilities and labels:'
zip(output_prob[top_inds], labels[top_inds])
得到的结果是:
[(0., 'n tabby, tabby cat'),#虎斑猫
(0.2379715, 'n tiger cat'),#虎猫
(0., 'n Egyptian cat'),#埃及猫
(0., 'n red fox, Vulpes vulpes'),#赤狐
(0., 'n lynx, catamount')]#猞猁,山猫
2.4 对比GPU
现在对比下GPU与CPU的性能表现
首先看看cpu每次(50 batch size)向前运行的时间:
%timeit net.forward()
%timeit能自动选择运行的次数 求平均运行时间,这里我的运行时间是1 loops, best of 3: 5.29 s per loop,官网的是1.42,差距
接下来看GPU的运行时间:
caffe.set_device(0)
caffe.set_mode_gpu()
net.forward()
%timeit net.forward()
1 loops, best of 3: 507 ms per loop(官网是70.2ms),慢了好多的说
2.5 查看中间输出
首先我们看下网络的结构及每层输出的shape,其形式应该是(batchsize,channeldim,height,weight)
# for each layer, show the output shape
for layer_name, blob in net.blobs.iteritems():
print layer_name + '\t' + str(blob.data.shape)
得到的结果如下:
data (50, 3, 227, 227) conv1 (50, 96, 55, 55) pool1 (50, 96, 27, 27) norm1 (50, 96, 27, 27) conv2 (50, 256, 27, 27) pool2 (50, 256, 13, 13) norm2 (50, 256, 13, 13) conv3 (50, 384, 13, 13) conv4 (50, 384, 13, 13) conv5 (50, 256, 13, 13) pool5 (50, 256, 6, 6) fc6 (50, 4096) fc7 (50, 4096) fc8 (50, 1000) prob (50, 1000)
现在看其参数的样子,函数为net.params,其中weight的样子应该是(output_channels,input_channels,filter_height,flier_width), biases的形状只有一维(output_channels,)
for layer_name,parame in net.params.iteritems():
print layer_name+'\t'+str(param[0].shape),str(param[1].data.shape)#可以看出param里0为weight1为biase
(96, 3, 11, 11) (96,)#输入3通道,输出96通道
(256, 48, 5, 5) (256,)#为什么变成48了?看下方解释
(384, 256, 3, 3) (384,)#这里的输入没变
(384, 192, 3, 3) (384,)
(256, 192, 3, 3) (256,)
fc6 () (4096,)#*3
fc7 () (4096,)
fc8 () (1000,)
可以看出只有卷基层和全连接层有参数
既然后了各个参数我们就初步解读下caffenet:
首先第一层conv1其输出结果的变化
(图片来自博客)
这一步应该可以理解,其权重的形式为(96, 3, 11, 11)
但是第二层的卷积层为什么为(256, 48, 5, 5),因为这里多了一个group选项,在cs231n里没有提及,这里的group=2,把输入输出分为了两个组也就是输入变成了96/2=48,
全连接层fc6的数据流图:
这是一张特拉维夫大学的ppt
下面进行可视化操作,首先要定义一个函数方便以后调用,可视化各层参数和结果:
def vis_square(data):
&&&Take an array of shape (n, height, width) or (n, height, width, 3) and visualize each (height, width) thing in a grid of size approx. sqrt(n) by sqrt(n)&&&
#输入为格式为数量,高,宽,(3),最终展示是在一个方形上
# normalize data for display
#首先将数据规则化
data = (data - data.min()) / (data.max() - data.min())
# force the number of filters to be square
n = int(np.ceil(np.sqrt(data.shape[0])))
#pad是补充的函数,paddign是每个纬度扩充的数量
padding = (((0, n ** 2 - data.shape[0]),
(0, 1), (0, 1))
# add some space between filters,间隔的大小
+ ((0, 0),) * (data.ndim - 3))
# don't pad the last dimension (if there is one)如果有3通道,要保持其不变
data = np.pad(data, padding, mode='constant', constant_values=0)
# pad with zero (black)这里该为了黑色,可以更容易看出最后一列中拓展的样子
# tile the filters into an image
data = data.reshape((n, n) + data.shape[1:]).transpose((0, 2, 1, 3) + tuple(range(4, data.ndim + 1)))
data = data.reshape((n * data.shape[1], n * data.shape[3]) + data.shape[4:])
plt.imshow(data)
plt.axis('off')
以conv1为例,探究如果reshape的
filters = net.params['conv1'][0].data
vis_square(filters.transpose(0, 2, 3, 1))
得到的结果
这里conv1的权重,原来的shape是(96, 3, 11, 11),其中输出为96层,每个filter的大小是11 11 3(注意后面的3噢),每个filter经过滑动窗口(卷积)得到一张output,一共得到96个。(下图是错误的,请去官网看正确的)
首先进入vissquare之前要transpose–》(96,11,11,3)
输入vis_square得到的padding是(0,4),(0,1),(0,1),(0,0) 也就是经过padding之后变为(100,12,12,3),这时的12多出了一个边框,第一个reshape(10,10,12,12,3),相当于原来100个图片一排变为矩阵式排列,然后又经过transpose(0,2,1,3,4)—&(10,12,10,12,3)又经过第二个reshape(120,120,3)
下面展示第一层filter输出的特征:
feat = net.blobs['conv1'].data[0, :36]#原输出为(50,96,55,55),这里取第一幅图前36张
vis_square(feat)
如果取全部的96张会出现下面的情况:中间的分割线没有了,为什么呢?
用上面的方法也可以查看其他几层的输出。
对于全连接层的输出需要用直方图的形式:
feat = net.blobs['fc6'].data[0]
plt.subplot(2, 1, 1)
plt.plot(feat.flat)
plt.subplot(2, 1, 2)
_ = plt.hist(feat.flat[feat.flat & 0], bins=100)#bin统计某一个数段之间的数量
输出分类结果:
feat = net.blobs['prob'].data[0]
plt.figure(figsize=(15, 3))
plt.plot(feat.flat)
大体就是这样了,我们可以用自己的图片来分类看看结果
主要分类过程代码主要步骤:
1. 载入工具包
2. 设置显示设置
3. 设置求解其set_mode_cup()/gpu()
4. 载入模型 net=caffe.Net(,,caffe.TEST)
5. transformer(包括载入均值)
6. 设置分类输入size(batch size等)
7. 载入图片并转换(io.load_image(‘path’), transformer.preprocesss)
8. net.blobs[‘data’],data[…]=transformed_image
9. 向前计算output=net.forward
10. output_prob=output[‘prob’][0]
11. 载入synset_words.txt(np.loadtxt(,,))
12. 分类结果输出 output_prob.argsort()[::-1][] ?????
13. 展示各层输出net.blobs.iteritems()
14. 展示各层参数net.params.iteritems()
15. 可视化注意pad和reshape,transpose的运用
16. net.params[‘name’][0].data
17. net.blobs[‘name’].data[0,:36]
18. net.blobs[‘prob’].data[0]#每个图片都有不同的输出所以后面加了个【0】
3 Fine-tuning
Now we will fine-tune the model we trained above on a different dataset to predict image style. we have 80000 images to train on. There will some changes :
1. we will change the name of the last layer form fc8 to fc8_flickr in our prototxt, it will begin training with random weights.
2. decrease base_lr andboost the lr_mult on the newly introduced layer.
3. set stepsize to a lower value. So the learning rate to go down faster
4. So in the solver.prototxt,we can find the base_lr is 0.001 from 0.01,and the stepsize is become to 20000 from 100000.
3.1 cmdcaffe
3.1.1 download dataset & model
we will only download 2000 images
python ./examples/finetune_flickr_style/assemble_data.py --workers=-1 --images=2000 --seed 831486
we have already download the model in the previous step
3.1.2 fine tune
let’s see some information in the new train_val.prototxt:
1. ImageData later
name:&data&
type:&ImageData&
transform_param{#预处理
mirror=true
crop_size:227#切割
mean_file:&yourpath.binaryproto&}
image_data_param{
batch_size:
new_height:
new_width: }}
另外加了一层规则化的。
在fc8_flickr层的lr_mult分别为10和20
./build/tools/caffe train -solver models/finetune_flick_style/solver.prototxt -weithts
models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel -gpu 0
3.2 pycaffe
some functions in python :
import tempfile
image=np.around()
image=np.require(image,dtype=np.uint8)
assert os.path.exists(weights)#声明,如果路径不存在会报错
在这一部分,通过ipython notebook定义了完整的网络与solver,并比较了微调模型与直接训练模型的差异,代码相对来说更加具体,由于下一边博客相关叙述比较仔细,这里就不重复了,但是还是很有必要按照官网来一遍的。
3.3 主要步骤
3.3.1 下载caffenet模型,下载Flickr数据
weights=’…..caffemodel’
3.3.2 defining and runing the nets
def caffenet():
n=caffe.NetSpec()
n.data=data
n.conv1,n.relu1=
n.drop6=fc7input=L.Dropout(n.relu6,in_place=True)
fc7input=n.relu6
if...else...
fc8=L.InnerProduct(fc8input,num_output=num_clsasses,param=learned_param)
n.__setattr__(classifier_name,fc8)#classifier_name='fc8_flickr'
if not train:
n.probs=L.Softmax(fc8)
if label is not None:
n.label=label
n.loss=L.SoftmaxWithLoss(fc8,n.label)
n.acc=L.Accuracy(fc8,n.label)
with tempfile.NamedTemporaryFile(delete=False)as f:
f.write(str(n.to_proto()))
return f.name
3.3.3 dummy data imagenet
L.DummyData(shape=dict(dim=[1,3,227,227]))
imagenet_net_filename=caffenet(data,train=False)
imagenet_net=caffe.Net(imagenet_net_filename,weights,caffe.TEST)
3.3.4 style_net
have the same architecture as CaffeNet,but with differences in the input and output:
def style_net(traih=True,Learn_all=False,subset=None):
if subset is None:
subset ='train' if train else 'test'
source='path/%s.txt'%subset
trainsfor_param=dict(mirror=train,crop_size=227,meanfile='path/xx.binaryproto')
style_data,style_label=L.ImageData(transform_param=,source=,batch_size=,new_height=,new_width=,ntop=2)
return caffenet(data=style_data,label=style_label,train=train,num_classes=20,classifier_name='fc8_filcker',learn_all=learn_all)
3.3.5 对比untrained_style_net,imagenet_net
3.3.6 training the style classifier
from caffe.proto import caffe_pb2
def solver():
s=caffe_pb2.SloverParameter()
s.train_net=train_net_path
if test_net_path is not None:
with temfile.Nxx as f:
f.write(str(s))
return f.name
bulit/tools/caffe train \ -solver models/path/sovler.prototxt\ -weights /path/.caffemodel\ gpu 0
def run_solvers():
for it in range(niter):
for name, s in solvers:
loss[][],acc[][]=(s.net.blobs[b].data.copy()for b in blobs)
if it % disp_interval==0 or it+1
...print ...
weight_dir=tempfile.mkdtemp()
weights={}
for name,s in solvers:
weights[name]=os.path.join(weight_dir,filename)
s.net.save(weights[name])
return loss,acc,weights
3.3.7 对比预训练效果
预训练多了一步:style_solver.net.copy_from(weights)
3.3.8 end-to-end finetuning for style
learn_all=Ture
本文主要分四部分 1. 在命令行进行训练 2. 使用pycaffe进行分类及特征可视化 3. 进行微调,将caffenet使用在图片风格的预测上 1 使用caffeNet训练自己的数据集 主要参考: 官方网址: http://ca
------分隔线----------------------------
相关阅读排行
相关最新文章
Copyright 2012- ( Coin163 ) All Rights Reserved &&为什么caffe训练网络时accuracy一直为0? - 知乎4被浏览1738分享邀请回答01 条评论分享收藏感谢收起caffe中做多标签训练,训练过程中accuracy一直不变? - 知乎26被浏览5701分享邀请回答0添加评论分享收藏感谢收起你的浏览器禁用了JavaScript, 请开启后刷新浏览器获得更好的体验!
Test net output #0: accuracy = 1
Test net output #1: loss = 0 (* 1 = 0 loss)
应该是过拟合了,你把基础学习率调小十倍试一下效果。
所以你的test集和train集是有重叠吗?
我也是被这个问题困扰了很久啊,我做图像分割的。name: &segnet&
name: &features&
type: &Data&
top: &features&
phase: TRAIN
transform_param {
data_param {
source: &project_shuo/lung_seg/train_db/features&
batch_size: 10
backend: LMDB
name: &labels&
type: &Data&
top: &labels&
phase: TRAIN
transform_param {
data_param {
source: &project_shuo/lung_seg/train_db/labels&
batch_size: 10
backend: LMDB
name: &features&
type: &Data&
top: &features&
phase: TEST
transform_param {
data_param {
source: &project_shuo/lung_seg/train_db/features&
batch_size: 10
backend: LMDB
name: &labels&
type: &Data&
top: &labels&
phase: TEST
transform_param {
data_param {
source: &project_shuo/lung_seg/train_db/labels&
batch_size: 10
backend: LMDB
name: &conv1&
type: &Convolution&
bottom: &features&
top: &conv1&
lr_mult: 1
decay_mult: 1
lr_mult: 2
decay_mult: 0
convolution_param {
num_output: 10
kernel_size: 7
name: &relu1&
type: &ReLU&
bottom: &conv1&
top: &conv1&
name: &conv2&
type: &Convolution&
bottom: &conv1&
top: &conv2&
lr_mult: 1
decay_mult: 1
lr_mult: 2
decay_mult: 0
convolution_param {
num_output: 1
kernel_size: 5
weight_filler {
type: &xavier&
bias_filler {
type: &constant&
name: &Sigmoid&
type: &Sigmoid&
bottom: &conv2&
top: &conv2&
name: &accuracy&
type: &Accuracy&
bottom: &conv2&
bottom: &labels&
top: &accuracy&
phase: TEST
name: &loss&
type: &SoftmaxWithLoss&
bottom: &conv2&
bottom: &labels&
top: &loss&
不知道怎么搞,数据集是从digits做的。运行的时候一直这样:
Waiting for data
I:03.79 data_layer.cpp:73] Restarting data prefetching from start.
I:03.78 data_layer.cpp:73] Restarting data prefetching from start.
I:11.74 solver.cpp:398]
Test net output #0: accuracy = 1
I:11.74 solver.cpp:398]
Test net output #1: loss = 0 (* 1 = 0 loss)
I:15.74 solver.cpp:219] Iteration 0 (-1.60881e-39 iter/s, 43.229s/50 iters), loss = 0
I:15.74 solver.cpp:238]
Train net output #0: loss = 0 (* 1 = 0 loss)
I:15.74 sgd_solver.cpp:105] Iteration 0, lr = 0.01
I:24.77 data_layer.cpp:73] Restarting data prefetching from start.
I:24.76 data_layer.cpp:73] Restarting data prefetching from start.
I:43.74 solver.cpp:331] Iteration 20, Testing net (#0)
I:13.79 data_layer.cpp:73] Restarting data prefetching from start.
I:13.78 data_layer.cpp:73] Restarting data prefetching from start.
I:20.74 solver.cpp:398]
Test net output #0: accuracy = 1
I:20.74 solver.cpp:398]
Test net output #1: loss = 0 (* 1 = 0 loss)
I:34.77 data_layer.cpp:73] Restarting data prefetching from start.
I:34.76 data_layer.cpp:73] Restarting data prefetching from start.
I:52.74 solver.cpp:331] Iteration 40, Testing net (#0)
I:23.79 data_layer.cpp:73] Restarting data prefetching from start.
I:23.78 data_layer.cpp:73] Restarting data prefetching from start.
I:30.74 solver.cpp:398]
Test net output #0: accuracy = 1
I:30.74 solver.cpp:398]
Test net output #1: loss = 0 (* 1 = 0 loss)
I:21.74 solver.cpp:219] Iteration 50 (0.163647 iter/s, 305.535s/50 iters), loss = 0
I:21.74 solver.cpp:238]
Train net output #0: loss = 0 (* 1 = 0 loss)
I:21.74 sgd_solver.cpp:105] Iteration 50, lr = 0.
I:44.77 data_layer.cpp:73] Restarting data prefetching from start.
I:44.76 data_layer.cpp:73] Restarting data prefetching from start.
I:02.74 solver.cpp:331] Iteration 60, Testing net (#0)
I:34.79 data_layer.cpp:73] Restarting data prefetching from start.
I:34.78 data_layer.cpp:73] Restarting data prefetching from start.
I:42.74 solver.cpp:398]
Test net output #0: accuracy = 1
I:42.74 solver.cpp:398]
Test net output #1: loss = 0 (* 1 = 0 loss)
请问楼主解决了吗
了吗,那个准确率一直是1的问题
一种可能是,你训练数据没有打散,即,负样本超级多后,大部分batch里全是负样本。
最好每个batch里即有负样本,又有一定比例的正样本.
顺便,还可以把训练了来的模型的参数值打出来看看,可能绝对直超大或0的值吧?
1. base_lr从0.01调小到0.001
2. train/val目录数据无重叠
3. train.txt和val.txt中数据都shuffle了
都没解决。
下面方法可行:
图片的分类标签label需要是int类型,caffe的convert_imageset工具/create_lmdb_train.sh脚本对应的train.txt/val.txt中每行的格式需要是&图片路径 分类标签的整数表示&的形式。
其中label = atoi(line.substr(pos + 1).c_str());
要回复问题请先或
浏览: 2079
关注: 5 人

我要回帖

更多关于 caffe accuracy layer 的文章

 

随机推荐