Watch 1 Star 1 Fork 1

回味一生 / HyperFT

Join us
Explore and code with more than 2 million developers,Free private repositories !:)
Sign up
This repository doesn't specify license. Without author's permission, this code is only for learning and cannot be used for other purposes.
移动端快速视频人脸跟踪-移动端150FPS+ http://www.zeusee.com spread retract

Clone or download
Cancel
Notice: Creating folder will generate an empty file .keep, because not support in Git
Loading...
README.md

HyperFT

Introduce

开源视频人脸跟踪算法,基于mtcnn人脸加测+onet人脸跟踪,移动端速度可以达到150fps+。该项目基于Android工程,提供底层JNI实现,使用者可以自行编译移植到其他平台。算法依赖ncnn深度学习计算库,体积小,易于集成。

代码编译

环境准备

Android Studio v3.5

CMake:3.6.4

Android SDK Platform-Tools:29.0.3

Android SDK Tools:26.1.1

NDK:r15c

依赖库

OpenCV:3.4.7

OpenCV

ncnn:20190611 bade132

ncnn

编译设置

1.设置Android NDK与Android SDK地址

2.修改CMake编译文件,在app/src/main/cpp/下修改CMakeLists.txt文件:

include_directories(D:/Wendell/Develop/libs/ncnn-android-lib/include)
include_directories(D:/Wendell/Develop/libs/OpenCV-android-sdk/sdk/native/jni/include)
set(OpenCV_DIR "D:/Wendell/Develop/libs/OpenCV-android-sdk/sdk/native/jni")
set_target_properties(libncnn PROPERTIES IMPORTED_LOCATION D:/Wendell/Develop/libs/ncnn-android-lib/${ANDROID_ABI}/libncnn.a)

相关资源

Demo(运行于 Ubuntu 18.04, i5-7400, NCNN)

image

体验

TODO

  • Sorry,目前还是需要依赖opencv,稍后提交精简版

  • iOS Project Develop

Comments ( 0 )

Sign in for post a comment

1
https://gitee.com/huiwei13/HyperFT.git
git@gitee.com:huiwei13/HyperFT.git
huiwei13
HyperFT
HyperFT
master

Help Search