Caffe学习系列(19): 绘制loss和accuracy曲线

如同前几篇的可视化，这里采用的也是jupyter notebook来进行曲线绘制。

 

In [1]:

#加载必要的库import numpy as npimport matplotlib.pyplot as plt%matplotlib inlineimport sys,os,caffe#设置当前目录caffe_root = '/home/bnu/caffe/' sys.path.insert(0, caffe_root + 'python')os.chdir(caffe_root)

 　　　　　　　　设置求解器，和c++/caffe一样，需要一个solver配置文件。

In [2]:

# set the solver prototxtcaffe.set_device(0)caffe.set_mode_gpu()solver = caffe.SGDSolver('examples/cifar10/cifar10_quick_solver.prototxt')

 　　　　　　如果不需要绘制曲线，只需要训练出一个caffemodel, 直接调用solver.solve()就可以了。如果要绘制曲线，就需要把迭代过程中的值

保存下来，因此不能直接调用solver.solve(), 需要迭代。在迭代过程中，每迭代200次测试一次

In [5]:

%%timeniter =4000test_interval = 200train_loss = np.zeros(niter)test_acc = np.zeros(int(np.ceil(niter / test_interval)))# the main solver loopfor it in range(niter):    solver.step(1)  # SGD by Caffe        # store the train loss    train_loss[it] = solver.net.blobs['loss'].data    solver.test_nets[0].forward(start='conv1')        if it % test_interval == 0:        acc=solver.test_nets[0].blobs['accuracy'].data        print 'Iteration', it, 'testing...','accuracy:',acc        test_acc[it // test_interval] = acc

 

Iteration 0 testing... accuracy: 0.10000000149Iteration 200 testing... accuracy: 0.419999986887Iteration 400 testing... accuracy: 0.479999989271Iteration 600 testing... accuracy: 0.540000021458Iteration 800 testing... accuracy: 0.620000004768Iteration 1000 testing... accuracy: 0.629999995232Iteration 1200 testing... accuracy: 0.649999976158Iteration 1400 testing... accuracy: 0.660000026226Iteration 1600 testing... accuracy: 0.660000026226Iteration 1800 testing... accuracy: 0.670000016689Iteration 2000 testing... accuracy: 0.709999978542Iteration 2200 testing... accuracy: 0.699999988079Iteration 2400 testing... accuracy: 0.75Iteration 2600 testing... accuracy: 0.740000009537Iteration 2800 testing... accuracy: 0.769999980927Iteration 3000 testing... accuracy: 0.75Iteration 3200 testing... accuracy: 0.699999988079Iteration 3400 testing... accuracy: 0.740000009537Iteration 3600 testing... accuracy: 0.72000002861Iteration 3800 testing... accuracy: 0.769999980927CPU times: user 41.7 s, sys: 54.2 s, total: 1min 35sWall time: 1min 18s

 　　　　　　绘制train过程中的loss曲线，和测试过程中的accuracy曲线。

In [6]:

print test_acc_, ax1 = plt.subplots()ax2 = ax1.twinx()ax1.plot(np.arange(niter), train_loss)ax2.plot(test_interval * np.arange(len(test_acc)), test_acc, 'r')ax1.set_xlabel('iteration')ax1.set_ylabel('train loss')ax2.set_ylabel('test accuracy')

 

[ 0.1         0.41999999  0.47999999  0.54000002  0.62        0.63  0.64999998  0.66000003  0.66000003  0.67000002  0.70999998  0.69999999  0.75        0.74000001  0.76999998  0.75        0.69999999  0.74000001  0.72000003  0.76999998]

Out[6]:

<matplotlib.text.Text at 0x7fd1297bfcd0>