Mood Detection using Facial Recognition Technology Application in Higher Education Institution

Abstract

Mood detection is an important trait that human can do using face expressions and it could be learned by the machine using machine learning approaches and technology. This research utilized a deep learning algorithm to develop a model that can detect various mood of participants in order to identify their subject areas of interest when enrolling in higher education institution so the academic advisor can smartly advise and guide the students in selecting the right program when they are seeking admission to the college.The methodology used in this research is application of the pre trained model technology deployed by convolution neural network for analyzing the facial expression on participants through three steps namely: image processing, facial recognition and mood detection. The mood detection by facial recognition technology is used to test independent input images of 15 participants to detect whether the mood is "happy", "sad", "neutral" and "surprise". The student's mood is determined by the facial movement this can further help in guiding the students for the proper program selection based on their facial expressions. A quasi experimental approach has been taken to analyze the results. The research findings show that the developed model was able to predict and classify various mood quiet accurately such as "happiness" and "neutral". On the other hand, difficultly detect other moods such as "surprise" and "sad".The study shows promising results and great capability of detecting human moods that has lots of applications such as academic advising in academic institution. The research is yet a good contribution in identifying a student's facial expression that positively effects the academic advising process and it is expected to better understand the student's needs and emotions.

Academic Emotion Classification Using FER: A Systematic Review

“…Based on the retrieved publications, Ayvaz et al [101], Candra Kirana et al [34], Dewan et al [35], El Hammoumi et al [31,32], Shi et al [42], Dash et al [45], Sabri [57], Murugappan et al [59], and Murugappan et al [60] have applied the Viola-Jones algorithm in the preprocessing step. Moreover, Ma et al [32], Yang et al [33], Hingu [62], and Zakka and Vadapalli [63] have used the Haar cascades method, while Hung et al [39], Lasri et al [40], and Alrayassi and Shilbayeh [50] have implemented the AdaBoost algorithm in the preprocessing phase.…”

“…However, there is room for improvement in this technique's speed, accuracy, and overall effectiveness with regard to applications. According to the studies conducted by Alrayassi and Shilbayeh [50], Principal Component Analysis (PCA) was used for the facial feature extraction. Besides feature extraction, dimensionality reduction is also included by applying the PCA algorithm.…”