Facial expression recognition for human-computer interaction

Tey, Yong Sheng (2025) Facial expression recognition for human-computer interaction. Final Year Project, UTAR.

PDF Download (4Mb)

Abstract

This project proposes the development of a robust Facial Expression Recognition (FER) system integrated within a chatbot framework, aimed at assisting individuals in recognizing and interpreting their emotional states more effectively. The system employs a deep learning-based approach, specifically utilizing the EfficientNet architecture, to improve the accuracy and reliability of emotion detection. It integrates advanced Convolutional Neural Network (CNN) techniques with preprocessing strategies to address variations in lighting conditions, patient demographics, and facial occlusions, thereby ensuring consistent performance across diverse clinical environments. Furthermore, the system delivers real-time feedback and personalized guidance based on the detected emotions, fostering more empathetic and patient-centered care. To achieve efficient real-time processing on resource-constrained medical devices, the model incorporates optimization techniques such as pruning, quantization, and lightweight CNN architectures. At the end, this project successfully trained a robust Facial Expression Recognition (FER) model with approximately 73% accuracy, integrated it with a chatbot system, and deployed the complete solution on a website to enable real-time, emotion-aware interactions.

Actions (login required)