Video based mobile application of emotion recognition by deep learning

OData support
Supervisor:
Dr. Szűcs Gábor
Department of Telecommunications and Media Informatics

The Artificial Intelligence also emerged as a new field of research in the middle of the 20th century. Recently, it has been very popular again, thanks to the development of technology. Whereas the large computing capacity enabled large amounts of data to be managed.

In the beginning of my thesis work I will present the machine learning, including the deep neural networks in more details, which are the subareas of the Artificial Intelligence. My work is not only about building the model, but also about introducing a field of application. My task was to create a system that can detect human emotions, where the user receives a feedback about their emotions via a mobile app.

The rest of my thesis described implementation of two more applications in details. First of all, there will be represented such a version, which is based on a server connection. The web server is capable of loading in pre-trained models and of recognizing emotions via videos, coming from mobile applications. I utilized Convolutional and Recurrent Neural Networks as models, in which a video database labelled with emotions meant the training data.

The other application of my thesis work is built on a new framework, which was released by Apple in September, 2017. With the help of it, the trained models can already be integrated into the application and are accessible directly from there.

Downloads

Please sign in to download the files of this thesis.