DCiFR

DCiFR (Demographic Characteristics in Facial Recognition) is a wrapper software that allows you to run deep learning models to parse demographic characteristics from an image. This open-source wrapper software written in Python has a GUI that will allow you to run complex models without any knowledge of coding. This includes functions from deepface and fairface and is built with PyQT5 to provide the GUI.

Getting Started

1. If you do not already have Python installed, navigate to this link to install it.

2. Download this repository to your local device.

3. In the command line, change your working directory to <download path>.

4. For initial run, enter the following in the command line.

• Windows

dcifr.sh

• Mac

sh dcifr.sh

5. After the initial run, use the following code to run DCiFR.

python3 dcifr.py

DeepFace Attributes

Based on faces within images, DCIFR's DeepFace pipeline reports results of four attributes: age, emotion, gender, and race.

• Age - Predicted age will fall between 0 - 100.
• Emotion - One of seven possible emotions: Angry, Disgust, Fear, Happy, Sad, Surprise, Neutral.
• Gender - Reports either man or woman.
• Race - The software predicts the probability of falling into one of seven race categories: Asian, black, Indian, Latino/Hispanic, Middle Eastern, or white. The results show the racial category with the highest probability.

More information on the attributes and how they are modeled can be found here.

FairFace Attributes

Based on faces within images, DCiFR's FairFace pipeline reports results of eight attributes: race, race4, gender, age, race_scores_fair, race_scores_fair_4, gender_scores_fair, and age_scores_fair.

• Race - Predicted probability of falling into one of seven race categories: White, Black, Latino_Hispanic, East Asian, Southeast Asian, Indian, or Middle Eastern.
• Race4 - Predicted probability of falling into one of four race categories: White, Black, Asian, or Indian.
• Gender - Reports either male or female.
• Age - Predicted age will fall within the following ranges: 0-2, 3-9, 10-19, 20-29, 30-39, 40-49, 50-59, 60-69, or 70+.
• Race_scores_fair - The model confidence score for predicting race.
• Race_scores_fair_4 - The model confidence score for predicting race4.
• Gender_scores_fair - The model confidence score for predicting gender.
• Age_scores_fair - The model confidence score for predicting age.

More information on the attributes and how they are modeled can be found here.

Output

The results will be saved in a DCIFR folder within the user's Documents as dcifr_Deepface_results or dcifr_Fairface_results with the date and time of creation attached to the end of the file name.

Reference

• deepface
• fairface
• PyQT5

License

DCiFR: Demographic Characteristics in Facial Recognition.
Copyright (C) 2021  Melina Raglin, Eungang (Peter) Choi, Erick Axxe

This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program.  If not, see <https://www.gnu.org/licenses/>.