Neural Network-Assisted Evaluation of Tea-Induced Emotional States

Tea consumption in Indonesia is associated with calming effects. Therefore, the objective of the present study was to quantitatively evaluate the relaxation potential of different tea samples by analyzing the induced alterations in facial expressions. Pearson’s correlation analysis revealed a moderate-to-strong positive association between total phenolic content and positive mood (r = 0.686, p < 0.01). Moreover, sex significantly affected most facial landmark measurements (p < 0.05), except for lip length, indicating distinct male and female facial geometry changes in response to tea consumption. Oolong tea elicited the strongest relaxation response, followed by green tea, with high-altitude Bali teas outperforming those from Banten (p < 0.05). Tea producers can leverage these findings to develop targeted relaxation beverages by selecting specific tea types and plantation elevation. Methodologically, total phenolic compounds were measured using the Folin–Ciocalteu assay. Sensory evaluation employed Sensomaker software to track emotional dynamics and dominant flavor sensations via Rate-All-That-Apply descriptors and a 15-point stress scale. An AI-powered tool was used for Facial Emotion Recognition and Landmark Detection to quantify subtle expression changes.