<![CDATA[Facial expressions play an important role as a non-verbal language rich in emotional information, especially in psychological contexts. The main challenge in this research is to understand and analyse complex facial expressions, which are often difficult for psychologists to interpret. The implementation of webcam-based facial expression recognition leverages the computer's ability to visually recognise human emotions, supported by artificial intelligence and machine learning. Convolutional Neural Network (CNN) and OpenCV methods are used to detect and classify facial expressions directly. The CNN model is trained using a dataset with six expression classes (happy, sad, angry, surprised, neutral, afraid), with four convolution layers for multi-class classification. The implementation of facial expression recognition is successful, the system captures facial images from a webcam, detects faces in the frame, and classifies facial expressions directly on the screen window. The performance of training data against the trained model measured using Classification Accuracy shows an accuracy of 72.34% in training accuracy and 60.54% in validation accuracy. While the performance of the facial expression recognition system calculated using Confusion Matrix resulted in an accuracy of 70.55%. The calculation results show that the model is at the Fair Classification parameter level or able to classify facial expressions in humans with a fairly good level of accuracy, this research has great potential for application development in the field of psychology. However, further optimisation is needed by involving experts to ensure its effectiveness.]]>
Copyrights © 2024
