tencent cloud

APIs

ドキュメントAPIsFace RecognitionIntroduction

Introduction

ダウンロード
フォーカスモード
フォントサイズ
最終更新日: 2026-10-20 17:15:43

Overview

Tencent Cloud Shentu · Face Recognition is based on Tencent YouTu's powerful face recognition and analysis technology. It provides various features including face detection and analysis, comparison, search, verification, facial feature positioning, and liveness detection. It offers developers and businesses a high-performance and high-availability face recognition service. Face Recognition has the following advantages:

  • Precise identification: Tencent Cloud face recognition service has set new records across multiple international open competitions. Face comparison achieved an accuracy of 99.80% in the 2017 LFW benchmark, and face search ranked first with a primary recognition rate of 83.29% in the million-scale MegaFace challenge, with industry-leading identification accuracy.
  • High stability: Tencent Cloud face recognition service has already passed validation with massive users and complex scenarios of Tencent internal products, ensuring stable operation and strong robustness, with service availability reaching 99.9%.
  • Leading algorithm: Based on the third-generation Tencent YouTu grandmother model, the model is optimized by integrating multiple training methods such as metric learning fusion, transfer learning, and multi-task learning. For the characteristics of different scenario-based businesses, customize finetuning or distilling models to meet the double requirements of business performance and latency.
  • Ease of use: provides a variety of APIs and Offline SDK for quick integration into applications and hardware devices, fully meeting cloud and device needs.
  • Diverse use cases: widely used in face recognition access control attendance, face sign-in, face recognition payment, face recognition login, and many other scenarios.
  • Real-time response: Face recognition features high concurrency, high throughput, and low latency. Even million-scale face search can be processed in just several hundred milliseconds, meeting your real-time use requirements.

All face recognition APIs introduced in this chapter comply with the latest OpenAPI 3.0 specification.
You can call APIs to perform operations on face recognition products, such as face detection and analysis, face comparison, face verification, facial feature localization, and liveness detection.
For all face recognition API information, see API overview.

Glossary

For common terms of the face recognition APIs, see the table below:

Terminology
 Description
Person library A person library is a collection of face libraries for known identities.
Liveness detection Detect a real-time recorded selfie video to confirm that the current user is a real person.
Dense key points Perform facial feature positioning (also known as facial keypoint localization) on the request image to obtain precise face info, return up to 888 key points, and precisely pinpoint the facial features and facial contour.
Static Liveness Detection High-Precision Version Compared with the existing static liveness detection service, the high-precision version enhances defense capabilities against attacks such as HD screens, cropped paper, and 3D masks while maintaining a high real person pass rate. Its attack interception rate is about 4-5 times that of similar types of products in the industry.
Face verification is used to judge whether a person is who they claim to be. The person's information is stored in the person library, and the person may contain several face images.
Face comparison Perform similarity comparison on human faces in two images and return the facial similarity score.

Usage Limits

For API parameter limits, refer to the parameter description in each API document.

Getting Started with APIs

You can use the API Explorer tool to call APIs online.
This document uses face detection and analysis in the face recognition product as an example. The steps to call the API through the API Explorer Tool are as follows:

  1. Register a Tencent Cloud account and verify identity. Then log in to the face recognition console, read and agree to the terms of service, click Enable Now to obtain the API interface call permission for face identity verification.
  2. Go to the API Explorer page. For more information about API Explorer tool usage, see Using API Explorer.
  3. Call the IAI API and select face detection and analysis.
  4. Manually input the corresponding parameters, make an online API call, and view the response result. For the input parameter description, please see the API Documentation (https://www.tencentcloud.com/document/product/867/44989?from_cn_redirect=1).
前の記事へ: History次の記事へ: API Category

ヘルプとサポート

この記事はお役に立ちましたか？

フィードバック

キーワードで検索
    Enter keywords to search documentation