Huawei's machine learning service live detection is upgraded again to build a safe and friendly interactive experience

Face recognition technology is widely used in public security, financial payment, traffic safety and other fields. The most frequently used scenario is that the user completes unlocking, payment and other actions through the face recognition technology on the smartphone, commonly known as "swiping face", and some developers will integrate face recognition technology in the application to facilitate the user to log in to the account and change the password. Wait.

However, while face recognition presents significant application value, the security risks of "fake faces" are gradually emerging. Currently, Huawei's machine learning service (ML Kit) has a relatively high daily life detection service, and its ability to distinguish the authenticity of other people's faces accurately meets the core demands of users, which has aroused the resonance of the majority of developers. In order to make face recognition more secure, we have added interactive live detection capabilities on the basis of silent live detection, hoping to build a safe and friendly face recognition experience with developers.

Living body detection makes "fake faces" invisible

In the impression of most people, face recognition technology is to allow machines to recognize people, but the current face recognition technology can only quickly identify the identity of a face image, but it cannot accurately distinguish the authenticity of other people's faces. So how to automatically and efficiently distinguish the authenticity of images and resist deception attacks to ensure system security has become an urgent problem in face recognition technology.

First of all, if we want to make human face recognition more secure, we need to detect fake and unreal faces-living body detection is the term used to refer to this algorithm. It is used to determine whether a human face is alive, including printed paper photos, electronic product display screens, silicone masks, stereoscopic 3D portraits, etc. Human faces presented with the help of other media can be defined as false and can resist all kinds of fake faces. attack.

Secondly, living body detection faces many challenges, and its rich applications in financial, public utilities, leisure and entertainment scenarios have brought some uncertainty. For example, different application scenarios have different performance requirements for living body detection. The diversity of equipment makes the performance of high, medium and low equipment vary greatly, as well as ethnic diversity and environmental changes. This requires the continuous change and update of live detection technology.

Newly upgraded interactive live detection to optimize user experience

In order to reduce the impact of the above uncertain factors, Huawei machine learning service (ML Kit) has added an interactive live detection capability. By adopting the method of command action coordination, the user can randomly choose 3 actions among the five actions of blinking, opening mouth, shaking the head left, shaking the right right, and watching. Mismatching instructions is considered to be forgery and deception. At the same time, guided detection is supported and identification scenarios are increased.

For scenes with occlusion and poor light, the interactive live detection capability uses deep learning models combined with image processing technology to accurately identify the detection scene and give guidance prompts. For example, prompts for faces that are too close or far away; prompts for dark and strong light; prompts for masks and sunglasses to block, etc., in order to achieve an accurate, efficient, safe and friendly humanized experience.

Nowadays, face recognition is closely related to our lives. Bank securities, financial insurance, people’s livelihood and social security, auto finance, housing rental, news media, etc. all have human-computer interaction scenarios. When users need to perform remote identity verification to confirm their identity information, The living body detection service can help users quickly achieve goals and tasks with a minimum of instructions, further reduce operating costs, and experience a fast and convenient facial identity verification process.

After the new upgrade, the live detection service will support both silent and interactive live detection methods.

• There are many breakthroughs in the algorithm of silent living detection. We have collected more than 200 types of data scenarios in cooperation with data companies to cover the diversity of user usage scenarios;
• Interactive live detection provides developers with a complete set of guidance controls and actual algorithm calling framework. Every developer can refer to the interactive UI for simple integration.

Companies can choose suitable live detection solutions to apply to various face recognition scenarios according to their own business needs. For example, various services such as insurance insured identity verification, real-name anti-addiction for game users, real-name card issuance by operators, live video broadcasting, and activation of rewarding authority.
The popularization of live detection services will continue to promote rapid technological innovation. In the future, Huawei's machine learning services will be based on AI technology and will continue to create live detection solutions with high security, high pass rate, and high ease of use for intelligent applications in various industries. To escort corporate risk control security and user personal information security.

For more details on bioassay services, please click:
https://developer.huawei.com/consumer/cn/doc/development/hiai-Guides/liveness-detection-0000001051386243#section2806658192714?ha_source=hms1

for more details>>

Visit Huawei Developer Alliance official website
Obtain development guide document
Huawei Mobile Services open source warehouse address: GitHub , Gitee

and learn about the latest technical information of HMS Core for the first time~