Skip to content

The Aim of this master thesis was to use saliency maps as a predictor for three emotional variables (Dominance, Arousal, Valence). The thesis is operated from the faculty for Psychological Methods with Interdisciplinary Focus, Goethe-University Frankfurt

License

Notifications You must be signed in to change notification settings

JNPauli/Emotional_Machines

Repository files navigation

Visual Saliency as a predictor for emotions - A Deep Learning Study

Jupyter Book Badge GitHub size GitHub issues GitHub PR License

Overview

This repository contains the code and documentation progress for my master thesis in the faculty for Psychological Methods with Interdisciplinary Focus. Goal of the Master Thesis was to train a Convolutional Neural Network with Saliency Maps created by following the approach of Simonyane et al., 2013 and classify three emotion variables Valence, Arousal and Dominance, see PAD-Theory

Structure

The open-lab-notebook holds information about the process of generating the research question.

The data folder holds the Dataset Exploration notebook. The code on how to load and process the data is stored in there. The Deep_Classification notebook is also hosted on google colab

Appendix

Please note that this project was conducted by a single person with only a bit of training in deep learning. My intrinsic goal was to improve my skillset and knowledge in deep learning.

About

The Aim of this master thesis was to use saliency maps as a predictor for three emotional variables (Dominance, Arousal, Valence). The thesis is operated from the faculty for Psychological Methods with Interdisciplinary Focus, Goethe-University Frankfurt

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published