Learned Features are better for Ethnicity Classification

This repository provides implementation for the paper, Learned Features are better for Ethnicity Classification. It is recommended to go through the Deep Face Recognition paper also.

Following are the softwares and tools needed to work with this repository.

  1. Python 3.5 or 3.6
  2. Matlab 2015 or above

Code presented here is tested with Python 3.6 and Matlab 2016

After installing Python and Matlab following are the packages that should be installed

  1. Python Packages (OpenCV, dlib, numpy)
  2. Matlab Toolbox(MatConvNet).

To install the MatConvNet follow the instructions given here

Download or clone the repository
Create a new folder named, VGG Network in the main porject direcotry
Download the pretrained VGG Face Network from here in VGG Network folder
Copy images into the Input_Images folder. Some example images are already included
Open the main.m in Matlab and run the code
Output class of image will be saved in result with image name and class name

Open source code snippets used in this project are Licensed. For details see Licenses folder

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
Classifier		Classifier
Input_Images		Input_Images
Licenses		Licenses
Reference		Reference
landmarks		landmarks
FaceCrop_224_224.m		FaceCrop_224_224.m
README.md		README.md
buildDetector.m		buildDetector.m
checkToolboxes.m		checkToolboxes.m
detectFaceParts.m		detectFaceParts.m
detectRotFaceParts.m		detectRotFaceParts.m
drawFourPoints.m		drawFourPoints.m
face_align.py		face_align.py
main.m		main.m
mergeFourPoints.m		mergeFourPoints.m

InzamamAnwar/Learned-Features-are-better-for-Ethnicity-Classification