Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/szad670401/hyperlandmark
Deep Learning Based Free Mobile Real-Time Face Landmark Detector. Contact:[email protected]
https://github.com/szad670401/hyperlandmark
106-points android cnn face-alignment face-landmark tracking
Last synced: 19 days ago
JSON representation
Deep Learning Based Free Mobile Real-Time Face Landmark Detector. Contact:[email protected]
- Host: GitHub
- URL: https://github.com/szad670401/hyperlandmark
- Owner: szad670401
- License: apache-2.0
- Created: 2018-07-03T09:00:05.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2020-09-01T05:43:26.000Z (about 4 years ago)
- Last Synced: 2024-05-21T02:28:58.731Z (6 months ago)
- Topics: 106-points, android, cnn, face-alignment, face-landmark, tracking
- Language: C++
- Homepage:
- Size: 92.9 MB
- Stars: 1,659
- Watchers: 96
- Forks: 524
- Open Issues: 32
-
Metadata Files:
- Readme: README.md
- License: LICENSE.md
Awesome Lists containing this project
README
# Free Mobile Real-Time Face Landmark 106 points Localization
### Introduce
良心级开源人脸标定算法,人脸美颜,美妆,配合式活体检测,人脸校准的预处理步骤.该项目Windows工程基于传统的SDM算法,通过修改开源代码,精简保留测试部分代码,优化代码结构.
Android代码基于深度学习,我们设计了高效的网络模型,该模型鲁棒性较好,支持多人脸跟踪.目前深度学习算法在人脸标定方向取得了良好的效果,该项目旨在提供一种较为简单易用的实现方式.#### Note
**2024.10.15**: 本项目**已经停止维护**,我们启动了一个**新的更加面向专业**的人脸相关开源项目。新的项目实现并集成了大部分SOTA的人脸相关算法,包括人脸识别、人脸检测、人脸(关键点)跟踪、人脸属性分析和活体检测等功能,具体项目地址可访问:[InspireFace](https://github.com/deepinsight/insightface/tree/master/cpp-package/inspireface)。
#### Related Resources
+ [基于HyperLandmark的活体检测](https://xiangzi.site/2019/04/19/iOS/%E6%B4%BB%E4%BD%93%E8%AF%86%E5%88%AB/)
+ [基于SDM的HyperLandmark-iOS实现](https://github.com/elhoangvu/HyperLandmark-iOS) (2019.02.06)
+ [相关技术博客](https://blog.csdn.net/lsy17096535/article/details/81116221)
+ [参考开源SDM算法](https://github.com/chengzhengxin/sdm)。
+ [基于CNN的人脸标定](https://github.com/lsy17096535/face-landmark)。
+ [HyperLandmark 106点序号](https://github.com/zeusees/HyperLandmark/blob/master/images/landmark_order.png)### Features
+ 106点,人脸轮廓描述更加细腻
+ 准确度高,逆光、暗光情况下依然可以取得良好的标定效果
+ 模型小,跟踪模型2MB左右,非常适合移动端集成
+ 速度快,Android平台代码在 Qualcomm 820 (st) 单张人脸7ms (2018.08.14)
+ 增加多人脸跟踪### APP
- 体验 Android APP:[http://demo.zeusee.com/ZeusFaceTrack](http://demo.zeusee.com/ZeusFaceTrack)
### Latest Update
+ Add OpenGL based camera interface.
### TODO
+ Dense face landmark (800 points)
+ Facial action recognition
+ Euler angle , face pose estimation.
+ Open source native tracking code.### Notes
Windows实现是基于免费intraface实现的,与android并不相同。
### 依赖
+ Windows demo 依赖 OpenCV
+ Android 可直接使用### Demo
![image](./resource/demo.gif)
![image](./resource/demo2.gif)
### Contact:
关于人脸标定我们正在进行数据收集与算法开发,技术交流、捐献数据可以加入下群。
+ HyperAI交流群 529385694### Author
- [Jack Yu]( https://github.com/szad670401)([email protected])