HAPPINESS MEASUREMENT SERVICE DEVELOPMENT BASED ON DEEP CONVOLUTIONAL NEURAL NETWORK

• Main Author: Andrian, Rizky
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/53459
• Institution: Institut Teknologi Bandung

One way to measure human emotions is through the recognition of human facial expressions. Research on human facial expression recognition has been carried out in recent years. The main focus in these studies is how to recognize human facial expressions from the characteristic features of human faces and then classify these expressions into several types of emotions. However, these studies resulted in an independent classification of human emotions between expressions. The total appearance of human facial expressions at a certain time cannot provide an interpretation that is in accordance with the actual emotional conditions. This was carried out in the Smart Happiness Meter study by Dilshad (2019) which produced measurement information based on the percentage of facial expressions over a certain time span. This study proposes an algorithm that combines facial expression recognition based on deep convolutional neural networks and psychological theory. The algorithm is implemented into the development of a web service-based prototype to measure the level of happiness based on the emotional intensity of facial expressions shown in a certain time span. The results showed that the proposed algorithm was able to provide a new level of happiness measurement interpretation according to the emotional state of facial expressions at a certain time span.