Face recognition application analysis

The application analysis of face recognition is mainly based on the two application interfaces of face comparison 1:1 and face search 1:n of Alibaba Cloud Vision Intelligence Open Platform, to explain various aspects.

1. Product
introduction Face comparison 1:1 : Input two pictures, compare them, and output the similarity of the two face pictures.

Face search 1:n : Input a single face picture, compare and search with the face pictures in the face database built by yourself on the platform, and output a face picture with high similarity.

 

2. Interface introduction:
face comparison 1:1
1. The input supports Aliyun ossURL and base64 input methods, and the input image quality score can be customized, which is used to intercept images with relatively poor input quality, so as to obtain from the source Control the quality of the image to make the result of face comparison more accurate.
2. Output the face coordinate frame point in the picture, the quality of the face picture, and the similarity between the two faces. How to define that two face pictures are the same person?
That is to judge the similarity of the face in the output result, for example: Confidence>=80. This value can be set according to the application scenario. If the security level is high, you can set a relatively higher threshold.

 

Face search 1:n
1. Face search 1:n searches based on the face database you have built, so the first thing you need to do is to build a face database, which is free.
2. Add face samples. I personally think that face samples are similar to a folder. Create a folder in the face library, and then store face pictures in the folder for distinction. This must be created. Only You can add face data only after creating a face sample.
3. Add face data, upload face pictures, and after uploading, the returned result is a string of face feature values, which are all face feature values ​​stored in the later face database, not face pictures.
4. Face search 1:n, based on the completion of the above steps, face search can be performed, and it supports million-level face database search. Still very fragrant. 

3. Application scenarios
Face comparison 1:1
1. APP login: use face comparison 1:1, you can log in with the personal account of the APP, upload a face picture when logging in, and store it in the database of the local APP Register face pictures, compare them, and complete login authentication.
2. Witness verification: Use face comparison 1:1 to complete the comparison between the on-site face picture and the face picture stored in the ID card chip. Currently, the most common application is to stay in a hotel.
Face search 1: n
1. Attendance system: Employees clock in, both mobile and PC can be used.
2. Face gate: 1:n face search, face recognition to open gates, communities, office building entrances and other scenarios.

You are welcome to refer to it. There will be more introductions about the application of Alibaba Cloud vision in the future. Welcome to communicate!

 

Guess you like

Origin blog.csdn.net/xiayexuyou/article/details/125635855