Implementing Face Recognition Login with Baidu AI and SpringBoot

Introduction

Motivated by the desire to create a personal face‑recognition demo, the author evaluated free APIs from several providers. After an unsuccessful attempt with Face++ due to an INVALID_OUTER_ID error, Baidu Cloud AI was chosen and successfully used to implement registration and login via facial recognition.

Difficulty Analysis

Both Face++ and Baidu Cloud AI expose REST APIs, but Face++ requires developers to extract and wrap the provided code snippets, demanding solid I/O stream handling and increasing code redundancy. Baidu Cloud AI offers two development paths: a manual code‑extraction approach similar to Face++, or a Maven‑based dependency that allows direct API calls, making the latter considerably easier.

Project Review (Baidu Cloud AI)

Final Effect Demonstration

Technology Stack

SpringBoot

Bootstrap

Thymeleaf

Baidu Cloud AI / Face++

Project Requirement Analysis

The goal is to replace manual username/password login with a face‑recognition flow: jQuery captures a photo from the local camera, the image is sent to a facial‑recognition API (Baidu Cloud or Face++), and the returned face data is used for automatic registration and subsequent login.

Project Setup

1. Preliminary Preparation

① Access the Baidu Cloud Face Recognition console (sign up with a phone number if you lack an account).

② Create a face‑recognition application, select the required interfaces (face detection, registration, search), and obtain the App ID, API Key, and Secret Key.

2. Test Baidu Cloud API

① Add the Maven dependency:

<dependency>
  <groupId>com.baidu.aip</groupId>
  <artifactId>java-sdk</artifactId>
  <version>4.9.0</version>
</dependency>

② Test the core APIs (face registration, detection, search) with the following Java snippets.

Face Registration

// Face registration example
@Test
public void testFaceRegister() throws IOException {
  AipFace client = new AipFace("AppId", "Api_key", "Api_secret");
  HashMap
map = new HashMap<>();
  map.put("quality_control","NORMAL");
  map.put("liveness_control","LOW");
  String path = "local_image_path";
  byte[] bytes = Files.readAllBytes(Paths.get(path));
  String encode = Base64Util.encode(bytes);
  JSONObject res = client.addUser(encode, "BASE64", "pdx", "1000", map);
  System.out.println(res.toString());
}

Face Detection

@Test
public void testFaceCheck() throws IOException {
  AipFace client = new AipFace("AppId", "Api_key", "Api_secret");
  String path = "local_image_path";
  byte[] bytes = Files.readAllBytes(Paths.get(path));
  String encode = Base64Util.encode(bytes);
  JSONObject res = client.detect(encode, "BASE64", null);
  System.out.println(res.toString(2));
}

Face Search

@Test
public void testFaceSearch() throws IOException {
  AipFace client = new AipFace("AppId", "Api_key", "Api_secret");
  String path = "local_image_path";
  byte[] bytes = Files.readAllBytes(Paths.get(path));
  String encode = Base64Util.encode(bytes);
  JSONObject res = client.search(encode, "BASE64", "pdx", null);
  System.out.println(res.toString(2));
}

Project Construction (Thymeleaf Template Engine)

1. Create Maven Project

Include the necessary dependencies and set up the directory structure.

2. Configure properties

ai.appId="api_id"
ai.apiKey="api_key"
ai.secretKey="api_secret"
ai.imageType=BASE64
ai.groupId="custom_group"
server.max-http-header-size=1000KB
spring.thymeleaf.cache=false

3. Wrap API Calls in a Utility Class

private AipFace client;
private HashMap
map = new HashMap<>();
public BaiduAiUtils(){
  map.put("quality_control","NORMAL");
  map.put("liveness_control","LOW");
}
@PostConstruct
public void init(){
  client = new AipFace(APP_ID, API_KEY, SECRET_KEY);
}
public Boolean faceRegister(String userId, String image){
  JSONObject res = client.addUser(image, IMAGE_TYPE, groupId, userId, map);
  return res.getInt("error_code") == 0;
}
public Boolean faceUpdate(String userId, String image){
  JSONObject res = client.updateUser(image, IMAGE_TYPE, groupId, userId, map);
  return res.getInt("error_code") == 0;
}

4. Front‑end Controller

@RequestMapping("/")
public String toLogin(){
  return "index";
}
/**
 * Face login endpoint
 */
@RequestMapping("/face-login")
@ResponseBody
public String searchface(@RequestBody @RequestParam(name="imagebast64") StringBuffer imagebast64, HttpServletRequest request) throws Exception {
  String userId = faceLoginService.loginByFace(imagebast64);
  request.getSession().setAttribute("userId", userId);
  request.getSession().setAttribute("username", "Patrick Star");
  return userId;
}

Important Notes Before Building

① If the QPS limit is reached, Baidu Cloud provides a free testing quota that can be claimed from the console.

② OAuth errors often stem from stray spaces in App ID, API Key, or Secret Key.

Project Summary

Using Baidu Cloud AI, the complete face‑recognition login system was assembled, achieving fast recognition and smooth user experience, though concurrent request handling could be improved. All source code (including the snippets above) is available in the public Gitee repository.

Repository: https://gitee.com/gao-wumao/ai-face-login