def batch_centerloss_confusionmatrix(imagepath="images", netpath):
#netpath="/home/g206/work_sdf/caffe1225/caffe/analysisfeatures/dvns/mnist"
# netpath="./dvns/mnist/"
modelist = [
"mnistorig_iter_10000.caffemodel",
"orimnistcaffe_alexnet_train_iter_400000.caffemodel",
"lenet_dvn__iter_500000.caffemodel",
"lenet_oridata__iter_500000.caffemodel",
"orimnitbvlc_googlenet_iter_400000.caffemodel",
"mnistgenerate_bvlc_googlenet_iter_480000.caffemodel",
]
deploylist = [
"alex_deploy.prototxt",
"alex_deploy.prototxt",
"lenet.prototxt",
"lenet.prototxt",
"google_deploy.prototxt",
"google_deploy.prototxt",
]
shapelists = [[10, 3, 227, 227], [10, 3, 227, 227], [64, 1, 28, 28],
[64, 1, 28, 28], [10, 3, 224, 224], [10, 3, 224, 224]]
netlayerlist = [
"fc8", "fc8", "ip2", "ip2", "loss3/classifier", "loss3/classifier"
]
outnamelist = [
"mnistlenetdvn.mat", "mnistlenetori.mat", "mnistalexdvn.mat",
"mnistalexori.mat", "mnistgoogledvn.mat", "mnistgoogleori.mat"
]
meanlist = [
"patchmnistimages_256_mean.binaryproto",
"patchmnistimages_256_mean.binaryproto",
"patchmnistimages_256_mean.binaryproto",
"patchmnistimages_256_mean.binaryproto",
"patchmnistimages_256_mean.binaryproto",
"patchmnistimages_256_mean.binaryproto"
]
from sklearn.metrics import confusion_matrix
y_true = []
#networklist=["lenet","alext","google"]
#net,transformer=load_models(modelpath = "./mninstmodel/lenet_iter_10000.caffemodel",deploypath = "./mnistmodel/lenet_deploy.prototxt",meanfacepath ="",shapelist=[64,1,28,28])
# net,transformer=load_models(modelpath = "/home/g206/work_sdf/caffe1225/caffe/analysisfeatures/dvns/mnist/orimnistcaffe_alexnet_train_iter_400000.caffemodel",deploypath = "/home/g206/work_sdf/caffe1225/caffe/analysisfeatures/dvns/mnist/alex_deploy.prototxt",meanfacepath ="",shapelist=shapelist)
#orimnitbvlc_googlenet_iter_400000.caffemodel
# net,transformer=load_models(modelpath = "/home/g206/work_sdf/caffe1225/caffe/analysisfeatures/dvns
imglist = GetFileList(imagepath, [])
i = 0
# /mnist/mnistgenerate_bvlc_googlenet_iter_480000.caffemodel",deploypath = "/home/g206/work_sdf/caffe1225/caffe/analysisfeatures/dvns/mnist/alex_deploy.prototxt",meanfacepath ="",shapelist=shapelist)
from tsne.tsne_1 import tsnepng
from mnist_single_plot_roc import drawroc
import traceback
for i in range(0, len(modelist)):
# for i in range(0, 2):
try:
pred = []
prediction = []
y_true = []
net, transformer = load_models(modelpath=netpath + modelist[i],
deploypath=netpath + deploylist[i],
meanfacepath=netpath + meanlist[i],
shapelist=shapelists[i])
y_true = mklabel(imglist, "mnist_test")
print "generating feature for model {0}....".format(modelist[i])
pred, prediction = generatefeature(imglist, net, transformer,
y_true, shapelists[i],
netlayerlist[i], outnamelist[i])
print len(y_true), len(pred)
print "roc" + outnamelist[i].split(".mat")[0] + ".png"
drawroc(y_true, prediction,
"roc" + outnamelist[i].split(".mat")[0] + ".png")
#tsnepng(prediction,y_true,"tsne_"+outnamelist[i].split(".mat")[0]+".png")
import pickle
with open(outnamelist[i].split(".")[0] + "pred.pkl", 'wb') as f:
pickle.dump(pred, f)
with open(outnamelist[i].split(".")[0] + "true.pkl", 'wb') as f:
pickle.dump(y_true, f)
y_true = np.asarray(y_true, dtype=int)
pred = np.asarray(pred, dtype=int)
print pred
print y_true
cm = confusion_matrix(pred, y_true)
print cm
with open(outnamelist[i].split(".")[0] + "_cm.pkl", "wb") as f:
pickle.dump(cm, f)
print cm
np.savetxt(outnamelist[i].split(".")[0] + "_cm" + ".csv",
cm,
delimiter=",")
except Exception as e:
print e
traceback.print_exc()
continue
mat.append(featline)
label = image.split("/")[-2]
# labels.append(str(lines[nn][1]))
labels.append(int(label))
# print "===>>",out['prob'].argmax()
# print "=====>>",lines[nn][1]
if (nn % 100 == 0):
with open("./error/png_oversmapling_result/" + outdir, 'w') as f:
scipy.io.savemat(f, {'data': mat, 'labels': labels}) # append
nn += 1
# print pred.shape
# tsnepng(mat,labels,"gootsne_"+outdir)
print "tsnepng=========================>>>>"
drawroc(labels, predroc, "./error/png_oversmapling_result/" + "zoomroc_10" + outdir.split('.')[0] + ".png")
print "roc=========================>>>>"
print(classification_report(labels, pred))
text_file = open("./error/png_oversmapling_result/" + outdir.split('.')[0] + ".txt", "w")
text_file.write(outdir.split('.')[0] + " model\n")
text_file.write(classification_report(labels, pred))
import pickle
with open("./error/png_oversmapling_result/" + outdir.split('.')[0] + "_pred.pkl", "wb") as f:
pickle.dump(mat, f)
with open("./error/png_oversmapling_result/" + outdir.split('.')[0] + "_true.pkl", "wb") as f:
pickle.dump(labels, f)
with open("./error/png_oversmapling_result/" + outdir, 'w') as f:
scipy.io.savemat(f, {'data': mat, 'labels': labels}) # append
cm = confusion_matrix(pred, labels)
with open("./error/png_oversmapling_result/" + outdir.split(".")[0] + ".pkl", "wb") as f:
pickle.dump(cm, f)
label = image.split("/")[-2]
# labels.append(str(lines[nn][1]))
labels.append(int(label))
# print "===>>",out['prob'].argmax()
# print "=====>>",lines[nn][1]
if (nn % 100 == 0):
with open("./error/mnist_result/" + outdir, 'w') as f:
scipy.io.savemat(f, {'data': mat, 'labels': labels}) #append
nn += 1
# print pred.shape
# tsnepng(mat,labels,"gootsne_"+outdir)
print "tsnepng=========================>>>>"
drawroc(
labels, predroc,
"./error/mnist_result/" + "zoomroc_10" + outdir.split('.')[0] + ".png")
print "roc=========================>>>>"
print(classification_report(labels, pred))
text_file = open("./error/mnist_result/" + outdir.split('.')[0] + ".txt",
"w")
text_file.write(outdir.split('.')[0] + " model\n")
text_file.write(classification_report(labels, pred))
import pickle
with open("./error/mnist_result/" + outdir.split('.')[0] + "_pred.pkl",
"wb") as f:
pickle.dump(mat, f)
with open("./error/mnist_result/" + outdir.split('.')[0] + "_true.pkl",
"wb") as f:
pickle.dump(labels, f)
with open("./error/mnist_result/" + outdir, 'w') as f:
