# node-seeta-legacy

**Repository Path**: Mitscherlich/node-seeta-legacy

## Basic Information

- **Project Name**: node-seeta-legacy
- **Description**: 👷 SeetaFaceEngine for Node.js.
- **Primary Language**: C++
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2022-03-04
- **Last Updated**: 2022-05-20

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# ![seeta logo](docs/seeta_logo.png)<br/>node-seeta

[![NPM version][npm-image]][npm-url]
[![npm download][download-image]][download-url]
[![MIT License][license-image]](License)

[npm-image]: https://img.shields.io/npm/v/seeta.svg?style=flat-square
[npm-url]: https://npmjs.org/package/seeta
[travis-image]: https://img.shields.io/travis/Mitscherlich/node-seeta.svg?style=flat-square
[travis-url]: https://travis-ci.org/Mitscherlich/node-seeta
[download-image]: https://img.shields.io/npm/dm/seeta.svg?style=flat-square
[download-url]: https://npmjs.org/package/seeta
[license-image]: https://img.shields.io/badge/License-MIT-blue.svg?longCache=true&style=flat-square

[SeetaFaceEngine](https://github.com/seetaface/SeetaFaceEngine2) 的 Node.js 插件.

* **[示例](#examples)**
* **[如何安装](#how-to-install)**
* **[使用说明](#usage)**

<a name="examples"></a>

# Examples

查看 [examples](examples) 了解更多详情.

<a name="how-to-install"></a>

# 如何安装

## 系统需求

|         | Ubuntu      | Windows    |
|:-------:|:-----------:|:----------:|
| 系统版本 | 16.04/18.04 | 7/8/8.1/10 |
| 体系结构 | x64         | x86/x86_64 |
| 最小配置 | 1 GB        | 1 GB       |
| 推荐配置 | 2+GB        | 2+GB       |

## 下载安装

* 源码地址: https://github.com/Mitscherlich/node-seeta.git
  * 克隆仓库 `git clone https://github.com/Mitscherlich/node-seeta.git` :hammer: [构建指南]](https://github.com/Mitscherlich/node-seeta/wiki/Build).

<a name="usage"></a>

# 使用说明

## 人脸检测

检测图片中的人脸边界盒位置。

```js
const FaceDetection = require('seeta').FaceDetection
const path = require('path')
// 1. 指定模型路径
const fdModel = path.join('path', 'to', 'models/SeetaFaceDetector2.0.ats')
// 2. 创建检测器对象
const detector = new FaceDetection(fdModel)
// 3. 检测图片中的人脸
const results = detector.detect(path.join('path', 'to', 'image/test.png'))
console.log({ result })
// 结果如下:
// {
//   "count": Number,     // 人脸个数
//   "data": {
//     "faces": [{
//       "x": Number,     // 人脸边界盒左上角 x
//       "y": Number,     // 人脸边界盒左上角 y
//       "width": Number, // 人脸边界盒宽度
//       "height": Number // 人脸边界盒高度
//     }]
//   }
// }
```

## 面部特征点定位

标记图像中制定面孔的特征点 (5 点)

```js
const { FaceDetection, PointDetector } = require('seeta')
const path = require('path')
// 1. 指定模型位置
const fdModel = path.join('path', 'to', 'models/SeetaFaceDetector2.0.ats')
const faModel = path.join('path', 'to', 'models/SeetaPointDetector2.0.pts5.ats')
// 2. 创建检测器对象
const detector = new FaceDetection(fdModel)
const pointer = new PointDetector(faModel)
// 3. 检测人脸位置并标定特征点
const faces = detector.detect(path.join('path', 'to', 'image/test.png'))
if (faces.data && face.data.faces.length) {
  const points = pointer.detect(path.join('path', 'to', 'image/test.png', faces.data.faces[0]))
  console.log({ result })
}
// 结果如下
// {
//   "data": {
//     "points": [{
//       "x": Number,     // 特征点 x
//       "y": Number,     // 特征点 y
//     }]
//   }
// }
```

## 人脸识别

1. `1 vs 1` 比较

比较两张图片中包含的人脸并得出相似度

```js
const { FaceDetector, PointDetector, FaceRecognizer } = require('seeta')
const path = require('path')
// 1. 指定模型路径
const fdModel = path.join('path', 'to', 'models/SeetaFaceDetector2.0.ats')
const faModel = path.join('path', 'to', 'models/SeetaPointDetector2.0.pts5.ats')
const frModel = path.join('path', 'to', 'models/SeetaFaceRecognizer2.0.ats')
// 2. 创建检测器对象
console.time('initailize')
const detector = new FaceDetector(fdModel)
const pointer = new PointDetector(faModel)
const recognizer = new FaceRecognizer(frModel)
console.timeEnd('initailize')
// 3. 检测图片中的面孔并标定特征点位置
const imagePath = join('path', 'to', 'data/test.png')

const faces = detector.detect(imagePath)
const face1 = faces.data.faces[0]
const face2 = faces.data.faces[1]

const points1 = pointer.detect(imagePath, face1).data.points
const points2 = pointer.detect(imagePath, face2).data.points
// 4. 比较两张图片中的面孔
const similar = recognizer.compare(
  { path: imagePath, points: points1 }, { path: imagePath, points: points1 }
)
console.log({ similar })
```

2. `1 vs n` 比较

- 第一步: 在数据库中注册图片

```js
const { FaceDetector, PointDetector, FaceRecognizer } = require('seeta')
const path = require('path')
// 1. 指定模型路径
const fdModel = path.join('path', 'to', 'models/SeetaFaceDetector2.0.ats')
const faModel = path.join('path', 'to', 'models/SeetaPointDetector2.0.pts5.ats')
const frModel = path.join('path', 'to', 'models/SeetaFaceRecognizer2.0.ats')
// 2. 创建检测器对象
console.time('initailize')
const detector = new FaceDetector(fdModel)
const pointer = new PointDetector(faModel)
const recognizer = new FaceRecognizer(frModel)
console.timeEnd('initailize')
// 3. 检测图片中的面孔并标定特征点位置
const imagePath = path.join('path', 'to', 'data/test.png')

const faces = detector.detect(imagePath)
const face = faces.data.faces[0]

const points = pointer.detect(imagePath, face1).data.points
// 4. 注册一张图片
const index = recognizer.register(imagePath, points)
console.log({ index }) // ==> the index in recognizer face database
```

- 第二步：比较图片相似度

```js
// 5. 检测一张新图片
const newImage = path.join('path', 'to', 'data/another_image.png')
const faces1 = detector.detect(newImage)
const face1 = faces1.data.faces[0]

const points1 = pointer.detect(newImage, face1).data.points
// 6. 在数据库中比较相似度
const results = recognizer.recognition(newImage, face1)
if (result.data && result.data.length) {
  console.log({ similars: result.data }) // ==> all similars in face databse
}
```

# License

[MIT](LICENSE)