python 导出 Caffe 结果并绘制 ROC 曲线

#!/usr/bin/env python
# -*- coding: UTF-8 -*-

import numpy as np
import matplotlib.pyplot as plt
#%matplotlib inline
import sys,os,caffe

################################################################################

caffe_root = '/home/zlf/my_software/my_caffe_configure/caffe-master/examples/ScatterParamsOld/result/GOOD/Accuracy3/'

sys.path.insert(0, caffe_root + 'python')
os.chdir(caffe_root)
if not os.path.isfile(caffe_root + 'cifar10_quick_iter_10000.caffemodel'):
    print "caffemodel is not exist..."
else:
    print "Ready to Go ..."
caffe.set_mode_gpu()

net = caffe.Net(caffe_root + 'deploy.protxt',
                caffe_root + 'cifar10_quick_iter_10000.caffemodel',
                caffe.TEST)

#print(net.blobs['data'].data.shape)


feat = [ [0 for i in range(3)] for j in range(96768)]
k = 0
right=0
imagePath = '/run/media/zlf/希捷娱乐/SSD_Caches/LungCancer/C/'
for filename in os.listdir(imagePath):              #listdir的参数是文件夹的路径
    print(filename[0:6])
    im = caffe.io.load_image(imagePath + filename)
    if filename[0:6] == 'LCN241' :
        label = 0
    else:
        label = 1    
    
    
    ################################################################################
    def convert_mean(binMean,npyMean):
        blob = caffe.proto.caffe_pb2.BlobProto()
        bin_mean = open(binMean, 'rb' ).read()
        blob.ParseFromString(bin_mean)
        arr = np.array( caffe.io.blobproto_to_array(blob) )
        npy_mean = arr[0]
        np.save(npyMean, npy_mean )
    binMean=caffe_root + 'mean.binaryproto'
    npyMean=caffe_root + 'mean.npy'
    convert_mean(binMean,npyMean)
    #print(np.load(npyMean))
    #print("np.load(npyMean).shape")
    #print(np.load(npyMean).shape)
    
    ################################################################################
    transformer = caffe.io.Transformer({'data': net.blobs['data'].data.shape})
    transformer.set_transpose('data', (2,0,1))
    transformer.set_mean('data', np.load(npyMean).mean(1).mean(1)) 
    transformer.set_raw_scale('data', 255)  
    transformer.set_channel_swap('data', (2,1,0))
    net.blobs['data'].data[...] = transformer.preprocess('data',im)
    inputData=net.blobs['data'].data
    
    ################################################################################
    net.forward()
    for layer_name, blob in net.blobs.iteritems():
        print layer_name + '\t' + str(blob.data.shape)
    
    
    feat12 = net.blobs['prob'].data[0].flatten()
    feat[k][0] = feat12[0]  #probability of normal 
    feat[k][1] = feat12[1]  #probability of cancer
    feat[k][2] = label      #original standard label
    aa = label == 1 & int(feat12[1] > feat12[0])
    bb = label == 0 & int(feat12[1] < feat12[0])
    if aa | bb :
        right = right + 1

    
    print('right' + str(right) )
    k = k + 1
    accu = right/k;
    print(k)
    print(filename[0:6])
    print(label)
    print( '       normal           cancer' )
    print(feat12)
    ################################################################################


from sklearn  import metrics
import math

y_true = [ x[2] for x in feat ] #标准的标签
 
y_scores = [ x[1] for x in feat ] #caffe输出的图片属于每一个类别的概率




fpr, tpr, thresholds = metrics.roc_curve(y_true, y_scores, pos_label=1)  

 


from sklearn.metrics import auc  

AUC = metrics.auc(fpr, tpr)  
rightsumaccuAUC = [right,k,accu,AUC]

np.savetxt(caffe_root + 'y_true.txt', y_true);
np.savetxt(caffe_root + 'y_scores.txt', y_scores);
np.savetxt(caffe_root + 'rightsumaccuAUC.txt', rightsumaccuAUC);


       
    
  
np.savetxt(caffe_root + 'fpr.txt', fpr);

np.savetxt(caffe_root + 'tpr.txt', tpr);

print('finish')

plt.plot(fpr, tpr)
plt.title('ROC_curve' + '(AUC: ' + str(AUC) + ')' )
plt.ylabel('True Positive Rate')
plt.xlabel('False Positive Rate')
plt.show()