Intel RealSense SR300 comparing detected face with a face from an external source

[OMore3]

I recently purchased the SR300 camera for a project I am working on. I need to save the recognized user in a database, so the app can load the data later for the camera to recognize the user as a registered user. I downloaded a sample application from the Intel website that allows users to register/unregister and save the registerd users to a database. The application is supposed to load the database and compare the detected face with a face stored in the loaded database, all featues in the sample application works, it detects users and register the user, but it nevers recognizes the user if the user is loaded from the database. I made sure all the drivers are up to date, and my computer meets the requirements. I also tried keeping the user's data in memory, reinitializing the camera, and set the user's data back to RecognitionConfiguration , and it didn't work. Can someone please help me?

[idata]

Hello gino.vaes,

Unfortunately, a method to store a file with the face recognition database doesn't exist .

And as I previously mentioned, there are no updates regarding the current status of that particular feature.

Have a nice day.

Regards,

Andres V.

[磊张3]

I have the same question.I purchased the SR300 camera for a project I am working on. I need to save the recognized user in a database, so the app can load the data later for the camera to recognize the user as a registered user. I downloaded a sample application from the Intel website that allows users to register/unregister and save the registerd users to a database. but, I want to know how to save the registerd users to a database ? Is this database a memory file or a real database (such as MySQL、oracle、sqlserver) ?

[MartyG]

The first thing I would recommend is to use the '2016 R2' SDK for face databases instead of '2016 R3 / Essentials' if you are not using R2 already. This is because R2 is the most stable release for face database creation.

Is this the sample you are using?

https://software.intel.com/en-us/articles/code-sample-facial-recognition-using-intel-realsense-sdk Code Sample: Facial Recognition Using Intel® RealSense™ SDK | Intel® Software

If so, please bear in mind with this sample that there are functions used, such as CreateStorage, that were not implemented in the RealSense SDK and so these lines should be 'commented out' to make the program ignore them, so that they do not cause problems. A section in the article describes about doing so.

In the sample, databases are saved as a text file that has bytes written to it. To quote from the sample article:

When a user clicks the Save Database button, a Byte array is declared and dimensioned to the size of the database, which is returned by calling the QueryDatabaseSize() method. The array is then populated by passing it to the QueryDatabaseBuffer() method. Finally, the database is committed to nonvolatile disk memory by simply calling File.WriteAllBytes().

Regarding mySQL database creation, a user who posted in 2015 said that it was possible by saving the data as blob data.

https://software.intel.com/en-us/forums/realsense/topic/595065# comment-1850183 Face Recognition Database

[磊张3]

Thank you for your quick repl y! I am sorroy my English is poor. I have purchased the sr300 camera, I want to use it for face recognition ,I have some questtion to ask you ! First,You say R2 is the most stable release for face database creation,I want to know can the 2016 R2' SDK use in the sr300 camera? Second,Do you provide two face contrast algorithms in your SDK（r2 or r3）? Third, Can you tell me how to implement the face recognition function use your sdk?

thanks!

[MartyG]

Don't worry, your English is good. But if you would find your own language easier, I can talk to you in that if you tell me which language you speak.

1. Yes, the SR300 works fine with the '2016 R2' SDK.

2. Yes, RealSense compares a face to a face image stored ("registered") in the database.

3. A lot of people who are using the Windows SDK for face recognition have used the sample program that I linked to. Near the top of the page is a 'download' button that says 'Code Sample zip' beside it. You can download the code for the sample with this button. If you get that program working and learn how the different parts of a face recognition system works then you can then change parts of the program for your own project.

[磊张3]

thank you ! I am in china ! Do I need to uninstall the sdksr3 installation Sr2, or do I download the sr2 Code Sample zip directly?

[MartyG]

I recommend uninstalling the R3 SDK and installing R2 (which is a 1.8 gb download) using the link below. Then download the Code Sample once R2 is set up.

You can download R2 in your browser using this link:

http://registrationcenter-download.intel.com/akdlm/irc_nas/vcp/9078/intel_rs_sdk_offline_package_10.0.26.0396.exe http://registrationcenter-download.intel.com/akdlm/irc_nas/vcp/9078/intel_rs_sdk_offline_package_10.0.26.0396.exe

***************

我建议卸载R3 SDK并使用下面的链接安装R2（这是一个1.8 gb的下载）。 一旦设置了R2，请下载代码示例。

您可以使用此链接在浏览器中下载R2：

[磊张3]

thank you ! I have installing R2 (which is a 1.8 gb download),but I have not found out about c # or javascript face detect examples of the sdk r2, can you tell me where to download the Code Sample R2 examples ? thanks!

[MartyG]

When you installed R2, it should have placed on your desktop a folder called Intel RealSense SDK Gold. Inside this folder is a Sample Browser application, where you can run samples that have already been built and also access the source code.

If you put the word 'face' into the Sample Browser's search box then you can find the C++, C# , Java and Javascript samples for Face Tracking.

I would recommend not using the Java and Javascript versions though, as the R2 SDK has a security vulnerability in its web component. Just use C# or C++.

***************

当您安装R2时，应该将桌面上放置一个名为Intel RealSense SDK Gold的文件夹。 在这个文件夹里面是一个样本浏览器应用程序，您可以在其中运行已经构建的样本，并访问源代码。

如果您将"face"一词放入Sample Browser的搜索框中，那么可以找到面部跟踪的C ++，C＃，Java和Javascript示例。

我建议不要使用Java和JavaScript版本，因为R2 SDK在其Web组件中具有安全漏洞。 只需使用C＃或C ++

[磊张3]

thanks!

I have successfully run RSSDK\\Samples\\DF_FaceTracking.cs demo,

but sometimes, click the start button, the camera turned on instantly automatically shut down, especially in the repeated operation of start, stop button, what reason is this excuse me?

[磊张3]

I think this process of face recognition should be like this:

1.to determine whether the camera face

2.from the database to extract data and the current camera face line matching

But I have several difficulties I don't know how to solve

1. the database is the face of the picture or face points?

2. the most important thing is that I don't know how to match Need to write their own algorithm comparison?

Can you help me? Thank you

[MartyG]

If you can get a successful face scan sometimes, and other times the camera disconnects, it sounds like the USB port is not reliably supplying the camera with the power it needs. Disconnection problems like this can be fixed in many cases by purchasing a mains-powered USB hub for around $15 and plugging the camera into that instead of directly into the PC. Powered hubs can be found on stores such as Amazon for around $15. As you are from China, if you wish to purchase one then I would recommend you do what I advised another Chinese user and buy it from JD.com. Search for 'powered usb 3.0 hub'.

http://www.jd.com/ 京东(JD.COM)-正品低价、品质保障、配送及时、轻松购物！

As databases are stored as a text file, it would be data that is stored, not a picture image.

The Intel documentation page for face recognition may be useful to you, as it has a script for accessing the database It is the second of the three scripts there, under the title 'Recognizing a face'. You query the database for whether a person's face is stored in it using the QueryUserID function.

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?doc_face_face_recognition_data.html Intel® RealSense™ SDK 2016 R2 Documentation

**********

有时您可以获得成功的脸部扫描，而有时候相机会断开连接，这听起来像USB端口不能可靠地为相机提供所需的电源。 像这样的断线问题在许多情况下可以通过购买主电源的USB集线器大约15美元，并将相机插入而不是直接插入PC中来解决。 有功能的集线器可以在亚马逊这样的商店上找到，价格在15美元左右。 正如你来自中国，如果你想购买一个，那么我建议你做我所建议的另一个中国用户，并从JD.com购买。 搜索'powered usb 3.0 hub'。

由于数据库作为文本文件存储，它将是存储的数据，而不是图片图像。

用于人脸识别的英特尔文档页面可能对您有用，因为它具有访问数据库的脚本。它是三个脚本中的第二个，标题为"识别面部"。 使用QueryUserID函数查询数据库中是否存储了一个人的脸部。

[磊张3]

thanks!

1. Does this SDK support local data file storage only and does not support connection databases (such as mysql, orcale)?

2. if you support the connection database, where is the connection information for the database configured?

[MartyG]

I found details of someone who used MySQL with a face recognition database.

https://software.intel.com/en-us/forums/realsense/topic/632767 Facial recognition on remote server

[磊张3]

thanks，

This example does not provide a way to store facial information into the database。

Now I am confused, that is if I will face information into byte stored in the database, when I get out, is not my own to write algorithms？ and it will slow, if the volume of data once, this will be a very long time

[MartyG]

There is no more information available about storing RealSense information in a MySQL database. So you will have to find a way to write something yourself if you want to do that. I do not have knowledge about doing that.

My understanding is that the RealSense face recognition database is quite fast if you are only using one database. RealSense can recognize 20 faces per database, and if there are more faces than that then they must be stored in further databases with 20 faces each in. So the more databases that you have, the slower the processing becomes as RealSense has to check through all the databases to find the identity of the person that it is seeing.

************

没有更多关于将RealSense信息存储在MySQL数据库中的信息。 所以如果你想这样做，你必须找到一种自己写点东西的方法。 我没有这方面的知识。

我的理解是，如果您只使用一个数据库，RealSense人脸识别数据库是相当快的。 RealSense可以识别每个数据库的20个面孔，如果有更多的面孔，那么它们必须存储在更多的数据库中，每个面都有20个面。因此，您拥有的数据库越多，处理变得越慢，因为RealSense必须检查所有 数据库找到它所看到的人的身份。

[磊张3]

More database, do you mean data local data files? A database can only save twenty faces, referring to a local data file can only save 20 faces? If i put all the faces in a table of Mysql, can you?

[MartyG]

I believe the 20 face limit for each database is set in the SDK software, not in the user's script. So whether you save your database as a text file or in a MySQL database, the RealSense SDK should only be able to check 20 faces in each database.

我相信每个数据库的20面值限制是在SDK软件中设置的，而不是在用户的脚本中。 所以无论您将数据库保存为文本文件还是将数据库保存在MySQL数据库中，RealSense SDK应该只能在每个数据库中检查20个面。

[磊张3]

If a database file can only recognize 20 faces, what should I do if there are lots of faces?

[MartyG]

If you have more than 20 faces, you need to create multiple database files with 20 faces each in them. Then RealSense checks all the database files until it finds a face that looks like the one the camera is seeing.

如果您有20个以上的面孔，则需要创建多个数据库文件，其中包含20个面。 然后RealSense会检查所有的数据库文件，直到找到一个看起来像相机看到的脸。

[磊张3]

Sdk seems to be the first to register, save the database to identify the face, if I have a picture of their own preservation (not saved by r3), can be directly through the realsense camera to identify it?