caffe学习笔记10 -- Fine-tuning a Pretrained Network for Style Recognitio

这是caffe官方文档Notebook Examples中的第四个例子，链接地址：http://nbviewer.jupyter.org/github/BVLC/caffe/blob/master/examples/03-fine-tuning.ipynb

这个实例用于在与训练好的网络上微调flickr_style数据。用已经训练好的caffe网络微调自己的数据。这种方法的好处在于，与训练网络从大量的图片数据集中学习而来，其中间层可以捕获一般视觉表现的“语义”, 可以将其看做一个包含强大特征的黑盒子，我们仅需要几层就能获得好的数据特征。

首先，我们需要保存数据，包含如下几步：

• 获取ImageNet ilsvrc与训练的模型
• 下载Flickr style数据集的一个子集
• 将下载的数据编译为caffe可以使用的格式

1. 导入程序需要的包：

import os
caffe_root = '/home/sindyz/caffe-master/'
os.chdir(caffe_root)
import sys
sys.path.insert(0,'./python')
 
import caffe
import numpy as np
from pylab import *

2. 导入数据和模型

# This downloads the ilsvrc auxiliary data (mean file, etc),
# and a subset of 2000 images for the style recognition task.
!data/ilsvrc12/get_ilsvrc_aux.sh
!scripts/download_model_binary.py models/bvlc_reference_caffenet
!python examples/finetune_flickr_style/assemble_data.py \
    --workers=-1 --images=2000 --seed=1701 --label=5

3. 比较一下两种模型的不同

!diff models/bvlc_reference_caffenet/train_val.prototxt models/finetune_flickr_style/train_val.prototxt  

输出这里省略

4. 用python学习，比较微调后的结果与直接训练的结果

niter = 200
# losses will also be stored in the log
train_loss = np.zeros(niter)
scratch_train_loss = np.zeros(niter)
 
caffe.set_device(0)
caffe.set_mode_gpu()
# We create a solver that fine-tunes from a previously trained network.
solver = caffe.SGDSolver('models/finetune_flickr_style/solver.prototxt')
solver.net.copy_from('models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel')
# For reference, we also create a solver that does no finetuning.
scratch_solver = caffe.SGDSolver('models/finetune_flickr_style/solver.prototxt')
 
# We run the solver for niter times, and record the training loss.
for it in range(niter):
    solver.step(1)  # SGD by Caffe
    scratch_solver.step(1)
    # store the train loss
    train_loss[it] = solver.net.blobs['loss'].data
    scratch_train_loss[it] = scratch_solver.net.blobs['loss'].data
    if it % 10 == 0:
        print 'iter %d, finetune_loss=%f, scratch_loss=%f' % (it, train_loss[it], scratch_train_loss[it])
print 'done'

输出省略。。。

5. 查看训练损失

可见，微调方法产生的损失波动平滑，而且比直接使用模型的损失小。

6. 将较小值部分放大：

plot(np.vstack([train_loss, scratch_train_loss]).clip(0, 4).T)

7. 查看经过200次迭代后，测试准确率，我们看到分类任务中有５个类别，随机测试的准确率为20%,与我们预期的一样，微调的结果要好于直接使用模型的结果。

test_iters = 10
accuracy = 0
scratch_accuracy = 0
for it in arange(test_iters):
    solver.test_nets[0].forward()
    accuracy += solver.test_nets[0].blobs['accuracy'].data
    scratch_solver.test_nets[0].forward()
    scratch_accuracy += scratch_solver.test_nets[0].blobs['accuracy'].data
accuracy /= test_iters
scratch_accuracy /= test_iters
print 'Accuracy for fine-tuning:', accuracy
print 'Accuracy for training from scratch:', scratch_accuracy

Accuracy for fine-tuning: 0.547999998927

Accuracy for training from scratch: 0.218000002205

 

参考资料：

http://nbviewer.jupyter.org/github/BVLC/caffe/blob/master/examples/03-fine-tuning.ipynb