hqli / face_recognition Goto Github PK
View Code? Open in Web Editor NEW使用webface人脸数据集以及DeepID网络,通过Caffe训练出模型参数,得到LFW二分类的人脸识别准确率。
使用webface人脸数据集以及DeepID网络,通过Caffe训练出模型参数,得到LFW二分类的人脸识别准确率。
我看很多人用python 的classifer函数做测试分类前会现将图片从rgb转为gbr通道,因为caffe训练时自动转过得。但是看你的代码中貌似没有处理?
你好,能不能说说你在什么卡上训练,大概跑了多久?
calculate_accuracy 里面的那个阈值设置怎么对结果没有丝毫影响,一开始设置的是0.20.8,后来我改成0.10.9,结果还是一样的?
def div_data(self, ratio):
#if num>select_ids[self.num-1][0]: should changed to next line
if num>select_id[0]:
做出来的识别率只有67%,想问一下是那个环节的问题。不知道您实现的识别率是多少,各项参数是否是按照上面代码的默认值
你好,还有一个问题,就是lmdb生成的时候,除了需要裁减之后的图片,对应的标签文件有什么?看了很久python源码也没有看出来。望解答,谢谢~
Hi, I also get a work to implement the DeepID recently, and I have searched many repositories about Face Verification and DeepID in github. I found none of them shows any features about sharing weights in their model configurations, which, however, is clearly included in 3th and 4th convolution layers of the model in Yi Sun's Deep Learning Face Representation from Predicting 10,000 Classes (http://mmlab.ie.cuhk.edu.hk/pdf/YiSun_CVPR14.pdf). I don't know why this happened. It seems like everyone ignored this characteristic for some conventional reason?
您好,我在运行您的程序最后一个环节人脸识别检验时候出现import cv2,出现Import Error:/Usr/local/lib/libopencv_gpu.so.2.4:undefined symbol:错误,一直没有解决掉,想请问您以下该怎么解决这个问题吗
首先是有大量图片读不出来:
E0522 08:28:06.982565 171153 io.cpp:80] Could not open or find file /home/ZengZhen/caffe-master/data/croped1/0000247/013.jpg 52
只读出少部分图片,我是用作者推荐的预处理工具进行处理的,然后最后报了如下错:
F0518 23:16:59.504396 140297 io.cpp:54] Check failed: fd != -1 (-1 vs. -1) File not found: /home/ZengZhen/caffe-master/examples/deepID/57/deepID_57_mean.binaryproto
麻烦作者了,不胜感激!
在切分训练集和测试集的时候,为什么加这个判断呢
if num>select_ids[self.num-1][0]: break
看原论文,说一个图片训练进行10000个分类,这个意思吗?还写了很多好处。这里不是很理解。
这里的人数num是不是也设定为10000?
请教一二,谢谢~~
你好,请问DeepID.py里面的'/media/ikode/Document/big_ma terials/document/deep_learning/caffe/face_datasets/webface/croped/'这个数据集可不可以分享一下!
想请问一下局部连接是怎么在caffe中实现的,如果可以,可以上传一下源码吗?
你好,请问你训练时的精度时多少以及loss是多少?我随机每个类选择两个样本做测试,其余作为训练集。训练精度只有0.46左右,而loss为3.1左右。能说说你的看法吗?表示感谢
RT
您好,我在运行代码的时候遇到了下面的问题:
Opened lmdb /home/dl/caffe-master/examples/deepID/20/deepID_20_train_lmdb
E0627 02:58:49.463901 11122 io.cpp:80] Could not open or find file /home/dl/caffe-master/data/20data/Abdel_Aziz_Al-Hakim/Abdel_Aziz_Al-Hakim_0001.jpg 12
E0627 02:58:49.463975 11122 io.cpp:80] Could not open or find file /home/dl/caffe-master/data/20data/Abdullah_al-Attiyah/Abdullah_al-Attiyah_0003.jpg 4
E0627 02:58:49.463984 11122 io.cpp:80] Could not open or find file /home/dl/caffe-master/data/20data/Abdullah_Ahmad_Badawi/Abdullah_Ahmad_Badawi_0001.jpg 9
以及
F0627 02:58:51.736857 11157 io.cpp:54] Check failed: fd != -1 (-1 vs. -1) File not found: /home/dl/li/LXW/caffe-master/examples/deepID/20/deepID_20_mean.binaryproto
*** Check failure stack trace: ***
的错误,我试了好多方法,不是路径问题,生成的lmdb文件不对,都是2 项,共 16.4 KB
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.