View Proposal

---

Proposer

John See

Title

ηEmo - Naturalistic Emotion Analysis Application for Web Video Streams

Goal

Design and development of a web stream based prototype application that performs analysis of the emotional state of person(s) in naturalistic scenarios

Description

Today's web streaming technology enables us to connect with others and interact with them remotely -- even interviews, work, and clinical assessments can now be conducted with the availability of video conferencing platforms. In this project, the aim is to build a prototype application (running in a web browser) that can analyse the emotional state of the person in the video stream. The definition of what it means by 'emotional state' can be further decided: categorical expressions (e.g. happy, sad), dimensional expressions (e.g. valence, arousal), and/or other related concepts like engagement levels or cognitive load. There are AI models available out there that can be directly utilised to achieve these tasks. The term "naturalistic" refers to real-world settings and environments where people express their emotions spontaneously rather than being instructed or told to act. *Depending on how the testing/verification process will be conducted, full ethical approval may be needed if people will be subjected to being assessed in the application.

Resources

A computer with a working camera. Possible package use: https://github.com/whitphx/streamlit-webrtc

Background

Reasonable level of competency in programming, especially Python; Some familiarity with porting machine learning models and API integration in apps would be an added advantage

Url

Difficulty Level

Moderate

Ethical Approval

Full

Number Of Students

1

Supervisor

John See

Keywords

emotion analysis, web application, stream processing

Degrees

Bachelor of Science in Computing Science