Model for Automatic Detection of the Big Five Personality Traits Through Facial Images
Keywords:
Affective computing, big five model, deep learning, machine learning, personality prediction, personality traitsAbstract
This paper presents a model based on deep learning for the automatic prediction of the big five personality traits: Extraversion, Agreeableness, Responsibility, Emotional Stability and Openness to Experience. For the development of the model a set of five neural networks was used, three of them built from an Inception Resnet V2 network, a network already trained to detect the facial area of a portrait provided by the DeepFace library and a FeedForward network built in this study using the results of the previous ones. To train these networks, a dataset of 13026 facial images was generated from a set of videos provided by Chalearn [1]. The model achieved an interesting accuracy rate, 64% was obtained by averaging the 5 factors (Extraversion = 64%, Agreeableness = 72%, Responsibility = 61%, Emotional Stability = 63% and Openness to Experience = 62%). Compared to other studies, the model presented in this article has several advantages: (1) it generated a new dataset for the study (2) it supports facial images with low quality, non-static and background and (3) the trained model can work with a large pixel size (299,299,3).
Downloads
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2023 International Journal of Computer Information Systems and Industrial Management Applications
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.