基于网络的人脸识别系统
我正在考虑建立一个基于网络的人脸识别系统。我知道有一些类似 KeyLemon 的软件，以及不同制造商提供的其他软件，允许笔记本电脑用户使用面部登录 Windows。我想知道这个功能是否可以转移到网络应用程序中。
I am thinking of building a web based face recognition system. I know there are a few like KeyLemon, and others offered by different manufacturers that allows the laptops users to login into Windows using their face. I am wondering if this functionality could be transfered to a web application.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（4）
建议您以此为基础
OpenCV（开源计算机视觉）是一个编程函数库实时计算机视觉。
Hacker Medley 上有一个关于 OpenCV 的精彩播客，其中包含各种有用的参考资料。据我了解，该库在开发方面往往发展得相当快，因此需要密切关注。
suggest you use this as the basis
OpenCV (Open Source Computer Vision) is a library of programming functions for real time computer vision.
There was an excellent podcast on OpenCV on Hacker Medley which has various references that are useful. From that i understand that the library tends to move quite fast in development terms so needs close attention.
你可以使用闪光灯之类的东西来访问相机...，然后使用相同的算法来识别脸部..
You may use thing like flash to access to camera ... , and then use the same algorithm to recognize the face ..
我写了一个网络应用程序，它做类似的事情。我不得不说 - 我对我们目前在此类事情上所达到的技术水平感到非常失望。该系统使用了 10mpix 佳能相机和特殊的手电筒支架。它必须有一个完美的白色背景，头部必须以正确的方式倾斜，不能旋转超过几度，并且必须与图片边缘有非常精确的距离。即便如此，它还是给出了很多误报和漏报。
所以也许他们今天想出了更好的办法，但我对此表示怀疑。这都是两年前的事了，该软件是一家专门从事此类事情的公司的商业产品。
所以总而言之，我说——最好不要。生物识别技术很酷，但目前它们太不稳定，只能在小众情况下部署。
I've written a web application which does something similar. And I have to say - I'm quite disappointed at the level of technology we're currently at for such things. The system in question used a 10mpix Canon camera and a special flashlight stand. It had to have a perfectly white background, the head had to be tilted exactly the right way, couldn't be rotated by more than a few degrees, and had to have very precise distances to the edges of the picture. And even then it gave a lot of false positives and negatives.
So maybe they've come up with something better today, but I doubt that. This was all 2 years ago and the software was some commercial product by a company that specializes in that sort of thing.
So all in all I say - better don't. Biometrics are cool, but currently they are way too unstable to be deployed in anything more than niche situations.
Keylemon 提供 Web API 来注册人脸并进行后续识别。您可以使用此 Web api 集成到您的应用程序中以提供人脸识别功能。它的工作原理是这样的。在注册期间，拍摄六张照片并生成生物识别模型。模型 ID 返回给客户端。该模型 ID 需要存储在应用程序数据库中。对于人脸识别，网络摄像头流与模型 ID 相结合被传递到 keylemon 服务器。如果模型 ID 和流匹配，则面部经过身份验证。
Keylemon provides web api to enroll faces and their later recognition. You can use this web apis to integrate in your application to provide face recognition functionality. It works like this. During enrollment six photographs are taken and a bio metric model is generated. A model id is returned to the client. This model id needs to be stored in application database. For face recognition, web camera streaming combined with model id is passed to the keylemon server. If the model id and stream matches the face is authenticated.