facenet-vs-vggface

file1	file2	facenet(欧式距离)	facenet(欧式距离-显卡)	vggface(余弦相似度)
gg1.jpg	gg2.jpg	0.6976	0.6545	0.74451
gg1.jpg	gg3.jpg	0.7641	0.6883	0.626138
gg1.jpg	gg5.jpg	0.7882	0.7961	0.58714
gg1.jpg	gg6.jpg	no detect	x	0.473986
gg1.jpg	gg7.jpg	no detect	x	0.475262
6.jpg	33.jpg	0.6209	x	0.467728
gg2.jpg	gg3.jpg	0.7289	0.7287	0.646415
g2.jpg	gg5.jpg	0.8284	0.8286	0.684934
gg2.jpg	gg6.jpg	no detect	x	0.560137
gg3.jpg	gg5.jpg	0.6176	0.6175	0.619669
rr.jpg	rr2.jpg	0.5816	x	0.545326

PS:facenet是基于tensorflow实现,vggface是基于caffe实现
还有TensorFlow的facenet运行相当慢，耗时是caffe的5倍

----------------------------------AI 学习路线图----------------------
1 先熟悉TensorFlow，caffe的环境，能安装，跑demo
2 从你感兴趣的课题切入，我比较感兴趣人脸识别，图像检测，觉得比较好玩，有兴趣才有动力嘛
然后学习下文本处理，比如分词，文本摘要生成，从github或开源的代码寻找相关感兴趣的，然后试着改成你想要的 4 自己尝试写一些力所能及的代码，体会其中的原理
5 逐步深入，优化开源的，加大难度

本来想以C++主要学习路线，caffe还可以全部用C++，TensorFlow就不行了，很多API只有python的

Name		Name	Last commit message	Last commit date
Latest commit History 96 Commits
picture		picture
10.jpg		10.jpg
11.jpg		11.jpg
12.jpg		12.jpg
13.jpg		13.jpg
16.jpg		16.jpg
17.jpg		17.jpg
18.jpg		18.jpg
19.jpg		19.jpg
2.jpg		2.jpg
20.jpg		20.jpg
201704051756.jpg		201704051756.jpg
21天实战caffe.txt		21天实战caffe.txt
3.jpg		3.jpg
6.jpg		6.jpg
8.jpg		8.jpg
9.jpg		9.jpg
A.csv		A.csv
B.csv		B.csv
BasicLSTMCell.py		BasicLSTMCell.py
C.csv		C.csv
ConfigMgr.cpp		ConfigMgr.cpp
ConfigMgr.h		ConfigMgr.h
HttpClient.cpp		HttpClient.cpp
HttpClient.h		HttpClient.h
README.md		README.md
SSIM.cpp		SSIM.cpp
Semphore.h		Semphore.h
StringUtil.h		StringUtil.h
TensorFlow：实战Google深度学习框架（第2版）配书代码及数据.txt		TensorFlow：实战Google深度学习框架（第2版）配书代码及数据.txt
auto_back.py		auto_back.py
clean.py		clean.py
cluster.py		cluster.py
cluster_gpu.py		cluster_gpu.py
cluster_monitoredTraining.py		cluster_monitoredTraining.py
cluster_monitoredTraining_gpu.py		cluster_monitoredTraining_gpu.py
cmp.py		cmp.py
compare_s.py		compare_s.py
conv1d.py		conv1d.py
conv1d_max_pool.py		conv1d_max_pool.py
db_model.py		db_model.py
dnnClassifier.py		dnnClassifier.py
dnnClassifier2.py		dnnClassifier2.py
douban.py		douban.py
dwxmly.py		dwxmly.py
epoll.cpp		epoll.cpp
epoll2.cpp		epoll2.cpp
expectscp		expectscp
extract-content.py		extract-content.py
f_fun.cpp		f_fun.cpp
feed.py		feed.py
file_util.py		file_util.py
format.csv		format.csv
format.py		format.py
get_image_baidu.py		get_image_baidu.py
get_variable.py		get_variable.py
gethome.php		gethome.php
hello.py		hello.py
infer		infer
infer_cnn		infer_cnn
install_dev_list.sh		install_dev_list.sh
international-airline-passengers.csv		international-airline-passengers.csv
ip_test.py		ip_test.py
ip_test_ui.py		ip_test_ui.py
iris.py		iris.py
iris.txt		iris.txt
iris_lr.py		iris_lr.py
iris_monitors.py		iris_monitors.py
iris_test.csv		iris_test.csv
iris_training.csv		iris_training.csv
isolation_forest.py		isolation_forest.py
label_encode.py		label_encode.py
line_regress.py		line_regress.py
linearRegressor.py		linearRegressor.py
loss_fun.py		loss_fun.py
lstm-binary-classify.py		lstm-binary-classify.py
lstm_keras.py		lstm_keras.py
matrix.py		matrix.py
mnist.py		mnist.py
monitoredTrainingSession.py		monitoredTrainingSession.py
monitoredTrainingSession_demo.py		monitoredTrainingSession_demo.py
net.py		net.py
night10w.csv		night10w.csv
np.py		np.py
optimizer.py		optimizer.py
optimizer1.py		optimizer1.py
pandas_example.py		pandas_example.py
process_manager.py		process_manager.py
queue_test.csv		queue_test.csv
r.txt		r.txt
relu.py		relu.py
sae.py		sae.py
seq2seq_inference.py		seq2seq_inference.py
session.py		session.py
shape.py		shape.py
sigmoid.py		sigmoid.py
sshlogin.sh		sshlogin.sh
stock_data.csv		stock_data.csv
tensorflow_compile		tensorflow_compile
test.py		test.py

berli/facenet-vs-vggface

Folders and files

Latest commit

History