cpuimage/mtcnn

MTCNN face detection implementation base on NCNN

Hey 👋🏽, I'm cpuimage

Hi, I am ZhiHan Gao, a native of Shantou, now living in Guangzhou, China.

I write open source projects about audio and image algorthms in github.

If you like my open source projects, and useful for you，

please consider buying me a coffee.

Thank you for your support!

Talking about Personal Stuffs:

• 👨🏽‍💻 I have worked for Baidu, KingSoft, etc.
• 🌱 I’m currently working on

  • Deep Learning

    • A Trimap-Free Solution for Real-Time Automatic Portrait Matting on Mobile Devices
    • A Robust Optimizer With Normalized Accelerated Convergence Capability in Deep Learning
    • A General and Adaptive Robust Loss Structure Scheme
    • A Robust Loss Weighting Solution For Learning Long-Tail Data
    • Image Synthesis and Semantic Manipulation Using Stable Diffusion Networks
    • Stable Diffusion Architecture Optimization And Deployment On Mobile Devices
    • A Robust Solution For Accelerated Training Convergence And Learning Long-Tail Data
    • A Arbitrary Resolution Super Resolution Solution for Real World
    • Accelerate Stable Diffusion FP16 Inference Deployment Optimization with TensorRT
    • Port Stable Diffusion X4 Upscaler To TensorFlow And Support FP16 Inference Deployment
    • Port Stable Diffusion PromptGen (GPT2) To TensorFlow And Support ONNX Inference Deployment
    • Improve Batch Normalization for Robust Training and Inference
    • Stable Diffusion Architectural Distillation
    • Content-aware 3-view synthesis based on Stable Diffusion in Game Art
    • Super Resolution Solution based on Stable Diffusion
    • Video Editing techniques based on Stable Diffusion
    • Port Stable Diffusion XL 1.0 To TensorFlow And Support FP16 Inference Deployment
    • A Plug-And-Play Algorithm For Asynchronous Inference With Frequency-Domain Decomposable Reconstruction For Arbitrary Visual Scenes
    • Stable Diffusion Inference With PyTorch Weights And More Features Like Stable Diffusion Web UI In Keras 3.x

  • Statistical Algorithms

    • Real time and embedded implementation of speech enhancement algorithms based on Minimum Mean-Square Error Short-Time Spectral Amplitude estimation (MMSE-STSA)
• 👯 I’m looking to collaborate on audio and image algorithms
  • 🤔 Reach me on
    • 
    • 
    • 
• 💬 Any paid technical service or solution consulting
  • 📫 Reach me on mail:
    • 

mtcnn's Issues

在ARM64下编译成功，但是测试结果完全错误

编译过程正常，但是测试结果和x86上有较大偏差，x86上测试结果正常，但是到ARM端编译出来的测试结果完全错位，而且检测人脸数量也不对。

可以支持gpu吗

请问程序可以支持gpu吗，项目原来是使用python+mxnet来做的，现在想提升效率，使用C++提升效率。

在generateBbox函数中，取score的时候，为什么取的是Score.channel[1]?

请教一下大佬，代码中generateBbox函数，score对应channel不是从0开始吗？为什么这里取score的时候从channel(1)？谢谢~

请问怎么转换成 tfserving

如题

target_link_libraries(mtcnn ${CMAKE_BINARY_DIR}/ncnn/src/libncnn.a m) issue

Dear Contributor
There are some issues, which i occur.
Issue 1: I have clone your this repo to local, and then use CMake to build a VS2013 project, it's all right,
but get "can't open libncnn.a library error", then i change the addition library path to fix this problem.
Issue 2: after fix issue , i get "can't open m.lib" so i remove this addition lib from my vs.
Issue 3: after fix all, i get some linked error. i copy the ncnn.lib from ncnn project output folder to this project, and then linked this lib to this project.
after those steps i build this repo to my PC, win7 VS2013.
so maybe this CMakeList have to make some improve.
Thanks very much your contribution!

编译出错。。

[ 90%] Built target caffe2ncnn
Makefile:129: recipe for target 'all' failed
make: *** [all] Error 2

"__cpuid”: 找不到标识符

/MTCNN/src/main.cpp:56:20: error: ‘browse’ was not declared in this scope

/home/tim/workspaces/AI_TEST_One_week/MTCNN/src/main.cpp:56:20: error: ‘browse’ was not declared in this scope
browse(saveFile);
^
CMakeFiles/mtcnn.dir/build.make:62: recipe for target 'CMakeFiles/mtcnn.dir/src/main.cpp.o' failed
make[2]: *** [CMakeFiles/mtcnn.dir/src/main.cpp.o] Error 1
CMakeFiles/Makefile2:72: recipe for target 'CMakeFiles/mtcnn.dir/all' failed
make[1]: *** [CMakeFiles/mtcnn.dir/all] Error 2
Makefile:129: recipe for target 'all' failed
make: *** [all] Error 2

Compiling help.

I compiled ncnn , how can I compile ?

it gave:

/Users/alpullu/Projects/ncnn/build/MTCNN/src/mtcnn.cpp:196:24: error: use of undeclared identifier 'ceil'

thx

在ARM v8下编译失败（raspberry pi 3b）

在支持neon的arm64系统下编译，layer.cpp处报错
“Pooling_arm_layer_creator” was not declared in this scope...
{"Pooling" , Pooling_arm_layer_creator},
在x86平台程序编译工作正常，在arm64下貌似找不到相应的层定义文件

simple cam example

Hi,

I am very new to ncnn.

can you add a very simple camera example? with opencv ?

Capture camera(Videocap (0)) detect faces / square and show with cv::im.show

I couldnt find a way to do this due to a newbee..

best

博主macos下面统译出错

404 行// float M_PI = 3.1415926535897932384626433832795f;
系统已经自定义了

87行，定义这个好像出错了
char drive[_MAX_DRIVE];
char dir[_MAX_DIR];
char fname[_MAX_FNAME];
char ext[_MAX_EXT];

下面是编译几个错误

/Users/hzq/demo/mtcnn/src/timing.h:31:5: error: non-void function 'nanotimer'
should return a value [-Wreturn-type]
return;
^
/Users/hzq/demo/mtcnn/src/main.cpp:87:13: error: use of undeclared identifier
'_MAX_DRIVE'
char drive[_MAX_DRIVE];
^
/Users/hzq/demo/mtcnn/src/main.cpp:88:11: error: use of undeclared identifier
'_MAX_DIR'
char dir[_MAX_DIR];
^
/Users/hzq/demo/mtcnn/src/main.cpp:89:13: error: use of undeclared identifier
'_MAX_FNAME'
char fname[_MAX_FNAME];
^
/Users/hzq/demo/mtcnn/src/main.cpp:90:11: error: use of undeclared identifier
'_MAX_EXT'
char ext[_MAX_EXT];

怎么编译和运行这个项目

博主，不好意思，我是个小白，请问一下这个项目怎么样可以正常运行，使用cmake还是写个makefile？能不能简单提供一下步骤？

次程序给出人脸相似度了吗

关于model文件bin和param

你的模型文件bin和param能不能由python生成的index、meta和data等文件转化来？如果可以，那过程又是怎么样？能说一下吗？谢谢！