szad670401 / hyperlandmark Goto Github PK
View Code? Open in Web Editor NEWDeep Learning Based Free Mobile Real-Time Face Landmark Detector. Contact:[email protected]
License: Apache License 2.0
Deep Learning Based Free Mobile Real-Time Face Landmark Detector. Contact:[email protected]
License: Apache License 2.0
Hi, it's cool! I wonder what's the training data you used to train the 106 points model?
如题所示,native代码和模型训练的开源计划?谢谢
用python导入发现是32位的so,改用32位python又出了一些问题,想请问还可以将so改为64位的吗?
如题。竖屏好好的,横屏就识别错乱了。在竖屏模式下,把头转90度也不行。
The Andriod Version is fantastic when I run it to my phone. In the futute ,is it possible to implement the IOS 's code?
Do you have plans to bring the Android implementation to Windows? It would be really cool to have a .dll or even the code, if it is possible.
Amazing work, by the way! The landmark tracking on the 106 point version is very nice and stable.
Hi team,
Thank for your awesome job. I am using it. In some cases, the landmark is not correct with face_confience rate is lower. The log message look like: I/jni_info: tracking_time:0.053353 face_confience:0.009775. I just don't know how can I get it on Java to avoid the case aforementioned.
基于Landmark,如何通过眨眼、打哈欠、左右、左顾右盼 来判定疲劳驾驶、安全驾驶呢?有没有计算公式或者demo 工程?
Hello Team,
Thanks a lot for the great work, but how to detect pose on android ?
I really need your help on that part
Thanks in advance
请问 安卓的代码人脸检测用的是MTCNN么,特征点检测算法用的是什么? 看工程里面貌似只有so和aar。
怎么改才可以让手机 前后摄像头都能识别人脸
如题,自己用ncnn打类似大小的模型大概只能用30fps。
android 工程, YUV 流 应该是没有做对应的旋转,导致后置摄像头识别不出来,
绘制人脸框和landmark点也是因为后置的屏幕尺寸有问题,导致位置出错。
望改善。
目前只提供预先编译的库。
可以简单说一下Android端模型的思路吗?
谢谢~
还有 faceBox[k].empty() 不对,我改为了faceBox[k].area() > 0
附加目录
E:\OpenCV\opencv\build\x86\vc14\bin
E:\OpenCV\opencv\build\include
链接器:
E:\OpenCV\opencv\build\x86\vc14\lib
依赖项:
opencv_calib3d2413.lib
opencv_calib3d2413d.lib
opencv_contrib2413.lib
opencv_contrib2413d.lib
opencv_core2413.lib
opencv_core2413d.lib
opencv_features2d2413.lib
opencv_features2d2413d.lib
opencv_flann2413.lib
opencv_flann2413d.lib
opencv_gpu2413.lib
opencv_gpu2413d.lib
opencv_highgui2413.lib
opencv_highgui2413d.lib
opencv_imgproc2413.lib
opencv_imgproc2413d.lib
opencv_legacy2413.lib
opencv_legacy2413d.lib
opencv_ml2413.lib
opencv_ml2413d.lib
opencv_nonfree2413.lib
opencv_nonfree2413d.lib
opencv_objdetect2413.lib
opencv_objdetect2413d.lib
opencv_ocl2413.lib
opencv_ocl2413d.lib
opencv_photo2413.lib
opencv_photo2413d.lib
opencv_stitching2413.lib
opencv_stitching2413d.lib
opencv_superres2413.lib
opencv_superres2413d.lib
opencv_video2413.lib
opencv_video2413d.lib
opencv_videostab2413.lib
opencv_videostab2413d.lib
Hi,
Thanks to share good work. How can I find the landmarks of image from gallery.
Thanks & regard
昨天准备试一下单张图片识别情况,我发现Android端是根据捕获视频流每帧数据识别的,我将图片转为bitmap格式传入后无法识别,后又将bitmap转为yue格式,还是无法识别,求解惑
你好,我这在选择后置摄像头后,用libyuv库函数对摄像头数据进行了镜像,但是还是无法识别人脸。问下,你那的镜像算法是什么??能否发给我下。谢谢
请问该代码及贵公司提供的demo app运行一段时间准时挂掉,是不是内存故意没有释放,还是偶然的bug存在?不是说好的开源吗,这让怎么用,也没有训练过程提供,这波广告做的好。
请问什么时候放出windows下106个特征点检测?当前windows下的68个特征点的检测,效果极为一般,没有任何使用意义,良心级安卓版本效果很好,为何不发布良心级win版本?
目前尝试了现有的kcf, enkcf,csk 等,在移动端都不理想
Is there a translated version into English?
你好,感谢你的分享。
我想请教点问题:
1、请问这个face detect 和 face alignment两个模型处理,中间是否使用了手机上的GPU?
2、如果没有使用手机上GPU,mtcnn的参数配置大概是什么样的?比如最小检测人脸多大?
3、face alignment 的模型使用的是博客上的squeezenetv2做backbone吗?
4、人脸检测有没有加跟踪的tricks?
添麻烦了
I got this when I first time install and run the app
2018-11-08 11:07:17.412 8340-8340/com.sample.multitrack106 W/Activity: Can request only one set of permissions at a time
2018-11-08 11:07:17.416 8340-8340/com.sample.multitrack106 D/mkdir: can't make folder
2018-11-08 11:07:17.417 8340-8340/com.sample.multitrack106 D/mkdir: can't make folder
2018-11-08 11:07:17.419 8340-8340/com.sample.multitrack106 W/System.err: java.io.FileNotFoundException: /storage/emulated/0/ZeuseesFaceTracking/models/det1.bin (Permission denied)
It seems that ActivityCompat.requestPermissions cannot be called twice. Should they be combined into one single call?
Hi, Thanks for your great job. Is there any plan to release android platform as open source? If yes, do you have any idea about date of release?
虽然速度很快,但是感觉不是很稳定啊,抖动很厉害,有什么解决方案吗
The windows project is a well known SDM implementation but i think your android implementation is different form windows one. What about algorithm and network structure and training phase?
Thanks.
识别过程中,能否获取到辅助参数,例如姿态角、脸部置信度这些信息呢?
谢谢你的分享。我还没有仔细看code,但我想问一下Prj-Win项目中有去抖动的操作吗?
楼主,传奇的106点我已经等了好久了,还没看到你提交呢
请问,《PFLD:A Practical Facial Landmark Detector》是你们的论文吗?
I try run Prj-Win, and I only get a 68 landmarks result. It is not 106 model. Where is the 106 model?
你好,我想问一下你的106个点的标注是用什么算法标注的,准确率大概有多高
前置摄像头开启,横屏检测不出人脸。希望横屏时亦能检测出人脸;如果在外面对图像进行旋转,需要消耗一定CPU,希望不增加消耗的情况下,实现横屏检测。
2,有时会有明显不是人脸被认为是人脸;
在关闭的issue中有提到,送入的图片必须旋转。所以横屏这一块应该是能实现的。
您好!
非常感谢开源这么棒的项目,请问如何量化评价106点landmark的好坏呢?
你好,请问一下,你们训练好这个模型,大约用了多少张图片呀?
我编译其他功能的时候需要 ncnn ,我加入了 libncnn.a 。编译没有问题。但是如果我同时加载这两个so 会崩溃
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.