引言

记录学习官网的例程中的一些重要语句，遇到的问题等，内容分散，建议顺序查看。
主要是调用Caffe的Python接口
源文件就在{caffe_root}/examples中(目录下面的中文标题也附有链接)，安装sudo pip install jupyter打开即可运行，初学者最好是放在它指定的目录，如，否则要改很多路径。
注：eaxmples是用jupyter notebook写的，部分Cell中出现了一些特殊的用法：
1. 感叹号‘！’：用于执行系统命令，如 !pwd
2. 百分号‘%’：用法太多，如 %matplotlib inline 显示绘图窗口 详见Jupyter Notebook Viewer

目录

用Python数据层在PASCAL上进行多标签分类

1. 多标签分类是多类别分类的泛化，其中每个实例（图像）属于多个类别。例如一幅图像既属于‘沙滩’类别，又属于‘度假图片类别’。在多类别分类中，一张图像仅属于单个类别。
2. Caffe通过SigmoidCrossEntropyLoss层支持多标签分类，下面使用Python数据层加载数据。 数据也可以通过HDF5或LMDB数据层提供，但python数据层提供了无限的灵活性。

1. 前期准备

1. 确保用了WITH_PYTHON_LAYER := 1编译Caffe
2. 下载PASCAL VOC 2012

3. 导入需要的模块

   import sys 
   import osimport numpy as np
   import os.path as osp
   import matplotlib.pyplot as pltfrom copy import copy% matplotlib inline
   plt.rcParams['figure.figsize'] = (6, 6)caffe_root = '../'  # this file is expected to be in {caffe_root}/examples
   sys.path.append(caffe_root + 'python')
   import caffe # If you get "No module named _caffe", either you have not built pycaffe or you have the wrong path.from caffe import layers as L, params as P # Shortcuts to define the net prototxt.sys.path.append("pycaffe/layers") # the datalayers we will use are in this directory.
   sys.path.append("pycaffe") # the tools file is in this folderimport tools #this contains some tools that we need

4. 设置数据集的路径并初始化Caffe

   
   # set data root directory, e.g:pascal_root = osp.join(caffe_root, 'data/pascal/VOC2012')# these are the PASCAL classes, we'll need them later.classes = np.asarray(['aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog', 'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor'])# make sure we have the caffenet weight downloaded.if not os.path.isfile(caffe_root + 'models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel'):print("Downloading pre-trained CaffeNet model...")!../scripts/download_model_binary.py ../models/bvlc_reference_caffenet# initialize caffe for gpu modecaffe.set_mode_gpu()
   caffe.set_device(0)

2. 定义网络prototxt

1. 用caffe.NetSpec 来定义网络，注意使用SigmoidCrossEntropyLoss层，用于多标签分类损失，还有Python数据层的定义！

   
   # helper function for common structuresdef conv_relu(bottom, ks, nout, stride=1, pad=0, group=1):conv = L.Convolution(bottom, kernel_size=ks, stride=stride,num_output=nout, pad=pad, group=group)return conv, L.ReLU(conv, in_place=True)# another helper functiondef fc_relu(bottom, nout):fc = L.InnerProduct(bottom, num_output=nout)return fc, L.ReLU(fc, in_place=True)# yet another helper functiondef max_pool(bottom, ks, stride=1):return L.Pooling(bottom, pool=P.Pooling.MAX, kernel_size=ks, stride=stride)# main netspec wrapperdef caffenet_multilabel(data_layer_params, datalayer):# setup the python data layer n = caffe.NetSpec()n.data, n.label = L.Python(module = 'pascal_multilabel_datalayers', layer = datalayer, ntop = 2, param_str=str(data_layer_params))# the net itselfn.conv1, n.relu1 = conv_relu(n.data, 11, 96, stride=4)n.pool1 = max_pool(n.relu1, 3, stride=2)n.norm1 = L.LRN(n.pool1, local_size=5, alpha=1e-4, beta=0.75)n.conv2, n.relu2 = conv_relu(n.norm1, 5, 256, pad=2, group=2)n.pool2 = max_pool(n.relu2, 3, stride=2)n.norm2 = L.LRN(n.pool2, local_size=5, alpha=1e-4, beta=0.75)n.conv3, n.relu3 = conv_relu(n.norm2, 3, 384, pad=1)n.conv4, n.relu4 = conv_relu(n.relu3, 3, 384, pad=1, group=2)n.conv5, n.relu5 = conv_relu(n.relu4, 3, 256, pad=1, group=2)n.pool5 = max_pool(n.relu5, 3, stride=2)n.fc6, n.relu6 = fc_relu(n.pool5, 4096)n.drop6 = L.Dropout(n.relu6, in_place=True)n.fc7, n.relu7 = fc_relu(n.drop6, 4096)n.drop7 = L.Dropout(n.relu7, in_place=True)n.score = L.InnerProduct(n.drop7, num_output=20)n.loss = L.SigmoidCrossEntropyLoss(n.score, n.label)return str(n.to_proto())

3. 创建网络和求解器文件，对于求解器，使用“tools”模块中的CaffeSolver类

• 该网络使用一个python数据层PascalMultilabelDataLayerSync，位于{caffe_root}/examples/pycaffe/layers/pascal_multilabel_datalayers.py，看看代码，非常简单，可以完全控制数据和标签。

• 下面和之前一样加载Caffe求解器

  solver = caffe.SGDSolver(osp.join(workdir, 'solver.prototxt'))
  solver.net.copy_from(caffe_root + 'models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel')
  solver.test_nets[0].share_with(solver.net)
  solver.step(1)

• 检查加载的数据

  transformer = tools.SimpleTransformer() # This is simply to add back the bias, re-shuffle the color channels to RGB, and so on...
  image_index = 0 # First image in the batch.
  plt.figure()
  plt.imshow(transformer.deprocess(copy(solver.net.blobs['data'].data[image_index, ...])))
  gtlist = solver.net.blobs['label'].data[image_index, ...].astype(np.int)
  plt.title('GT: {}'.format(classes[np.where(gtlist)]))
  plt.axis('off');

• 补充：简单介绍一下{caffe_root}/examples/pycaffe/tools.py，主要是定义了两个类SimpleTransformer和CaffeSolver
  
  1. SimpleTransformer是为Caffe预处理和逆向处理图像，总共有四种方法分别为：set_mean、set_scale、preprocess、deprocess，其中预处理包括：转数据类型、改通道(RGB->BGR)、减均值、调比例、改维度(HWC->CHW)，逆向处理则是相反操作！
  2. CaffeSolver是一个用于创建solver.prototxt文件的类，它设置了默认值并可以导出求解器参数文件，主要有两个方法：add_from_file用于读取prototxt文件并更新实例参数；write用于导出求解器参数(按字母排序)到指定路径。

4. 训练网络

• 首先需要一些方法来测试精度，汉明距离（Hamming distance）是多标签问题中广泛使用的，写一个简单的测试循环如下：

  def hamming_distance(gt, est):return sum([1 for (g, e) in zip(gt, est) if g == e]) / float(len(gt))def check_accuracy(net, num_batches, batch_size = 128):acc = 0.0for t in range(num_batches):net.forward()gts = net.blobs['label'].dataests = net.blobs['score'].data > 0for gt, est in zip(gts, ests): #for each ground truth and estimated label vectoracc += hamming_distance(gt, est)return acc / (num_batches * batch_size)

• 都准备好了，开始训练一段时间

  for itt in range(6):solver.step(100)print 'itt:{:3d}'.format((itt + 1) * 100), 'accuracy:{0:.4f}'.format(check_accuracy(solver.test_nets[0], 50))

• 检查输出结果，看起来精度在增加，似乎收敛很快，但是一开始精度就这么高有些不正常，原因可能是 PASCAL数据集只有20类，而每张图片通常只有1-2类，GT很稀疏，因此如果预测全为零会产生很高的准确度，下面来检查一下是不是这样：

  def check_baseline_accuracy(net, num_batches, batch_size = 128):acc = 0.0for t in range(num_batches):net.forward()gts = net.blobs['label'].dataests = np.zeros((batch_size, len(gts)))for gt, est in zip(gts, ests): #for each ground truth and estimated label vectoracc += hamming_distance(gt, est)return acc / (num_batches * batch_size)print 'Baseline accuracy:{0:.4f}'.format(check_baseline_accuracy(solver.test_nets[0], 5823/128))

6. 查看一些预测结果

test_net = solver.test_nets[0]
for image_index in range(5):plt.figure()plt.imshow(transformer.deprocess(copy(test_net.blobs['data'].data[image_index, ...])))gtlist = test_net.blobs['label'].data[image_index, ...].astype(np.int)estlist = test_net.blobs['score'].data[image_index, ...] > 0plt.title('GT: {} \n EST: {}'.format(classes[np.where(gtlist)], classes[np.where(estlist)]))plt.axis('off')

上一篇：Caffe实战之Python接口系列（四）Brewing Logistic Regression then Going Deeper

下一篇：Caffe实战之Python接口系列（六）Net Surgery(Editing model parameters)