search

manassikri / Liveliness-Detection

This repository contains the code and all the pre trained files which are used for liveliness detection of human facing for building an anti spoofing algorithm
4 stars 0 forks source link

关于caffemodel #1

Open jiagh2010 opened 5 years ago

jiagh2010 commented 5 years ago

这个模型做什么用，我看代码用dlib进行人脸检测定位，另外一个caffe模型用作什么？最后代码看着是统计眨眼次数？？眨眼了就是live？？

jiagh2010 commented 5 years ago

What is the Caffe model used for? @manassikri

manassikri commented 5 years ago

What is the Caffe model used for? @manassikri

We are not basically training a network to detect human faces, but are using a pre-trained model which is in caffe form. Basically caffe models are pre trained models which can be used for a variety of tasks from using them in simple visualization techniques, regressions, robotics etc. Here we used for face detection.

To use that model we use the following command net = cv2.dnn.readNetFromCaffe(args["prototxt"], args["model"])

Here we have given to caffe models 1)protxt 2)model

The parameters are defined in prototxt file while the trained cnn model can be found in model.

For further understanding read the following links:-

https://shengshuyang.github.io/A-step-by-step-guide-to-Caffe.html

http://adilmoujahid.com/posts/2016/06/introduction-deep-learning-python-caffe/