Αναγνώριση συναισθημάτων μέσω εικόνων με την χρήση συνελικτικών νευρωνικών δικτύων
Emotion recognition through images using convolutional neural networks

View/ Open
Keywords
Συνελικτικά νευρωνικά δίκτυαAbstract
The present work focuses on the way neural networks, specifically Convolutional Neural Networks, recognize various emotions through a specific dataset of images. The first chapter provides an introduction to the subject of the work, while also formulating the problem and analyzing its significance (Analysis of the concept of 'Emotion'). In the second chapter, the concept of 'Affective Computing' is analyzed, as well as the relationship between machine learning and the concept of emotion, and the correlation between 'Affective Computing' and neural networks. In Chapter 3, the term 'Convolutional Neural Networks' is introduced. The differences between 'Fully Connected Neural Networks (FNNs)' and 'Convolutional Neural Networks (CNNs)' are analyzed, why Convolutional Neural Networks are the most suitable when our dataset consists of images, as well as the main characteristics of CNNs. In chapter 4, the most basic and widespread 'Image Datasets' are examined, the process of image processing, the concept of 'Data Augmentation', as well as the different ways we address 'Class Imbalance' in our dataset. In chapter 5, the Experimental Design of the Project I have created and the Experimental Results of this Project are explained. In the final chapter, chapter 7, I draw conclusions, suggest future extensions/improvements, and present final observations. Finally, the Bibliography is appended.


