Skip to content

Latest commit

 

History

History
79 lines (69 loc) · 2.69 KB

README.md

File metadata and controls

79 lines (69 loc) · 2.69 KB

Source codes for the ChaLearn Looking at People - Job Candidate Screening Competition

Info: The code is not optimized yet ..

Scientific Use

If you find thiw work useful please cite the following documents:

@InProceedings{CVPRW2017,
title={Personality Traits and Job Candidate Screening via Analyzing Facial Videos},
author={Bekhouche, Salah Eddine and Dornaika, Fadi and Ouafi, Abdelkrim and Taleb-Ahmed, Abdelmalik},
booktitle={Computer Vision and Pattern Recognition Workshops (CVPRW), 2017 IEEE Conference on},
pages={1660–1663},
year={2017},
organization={IEEE},
doi={10.1109/CVPRW.2017.211}
}

AND

@article{ESWA2017,
title = “Pyramid Multi-Level Features for Facial Demographic Estimation “,
journal = “Expert Systems with Applications “,
volume = “80”,
pages = “297–310”,
year = “2017”,
issn = “0957-4174”,
doi = “http://dx.doi.org/10.1016/j.eswa.2017.03.030”,
url = “http://www.sciencedirect.com/science/article/pii/S0957417417301793”,
author = “SE. Bekhouche and A. Ouafi and F. Dornaika and A. Taleb-Ahmed and A. Hadid”
}

Requirements

  • Operating System: Ubuntu or Debian
  • Matlab R2016a or later
  • OpenCV
  • Dlib (using GCC5)
  • Python2.7

Setup

Download the project

$ git clone https://github.com/Bekhouche/CVPR2017.git

Download and prepare data

You have to execute this bash script (database.sh) in Terminal, so you can type this instruction in Terminal:

$ ./database.sh

In case you have a problem with permission; you have to execute this command in terminal:

$ sudo chmod -R 777 database.sh

Face preprocessing phase

Same as above, execute this bash script (face_preprocessing.sh)

$ ./face_preprocessing.sh

Feature extraction phase

Execute this bash script (feature_extraction.sh)

$ ./feature_extraction.sh

In case you want to skip the last two phases. Download the features and the relative files from Mendeley or Researchgate. You have to put database.mat in database folder, for the features files download PML_BSIF_7.mat and PML_LPQ_7.mat put them in feature_extraction/features path.

Personality traits estimation phase

Execute this bash script (personality_traits_estimation.sh)

$ ./personality_traits_estimation.sh

After running the script, you will get "predictions.pkl". This file can be used to get the scores.

Licence

The PML-FirstImpressionsV2 code is licensed under the MIT License.