Intelligent Multimodal System for the Early Detection of Children's Needs and Emotions through Facial Expression Analysis

Garcia H.; Cerna S.; Mendoza C.; Llanos J.; Bazán L. · Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology (2026) · DOI: 10.18687/laccei2026.1.1.1217

Resumen

At present, verbal communication between infants and their caregivers constitutes a clinical and everyday challenge, since many first-time parents and daycare staff often make ambiguous interpretations of the signals used by babies to express their needs. The objective of this research was to develop an intelligent system capable of detecting and classifying infants' emotions in real time through the analysis of their gestures and facial expressions, as well as acoustic signals (crying), implemented and evaluated at the childcare center “El Nido.” The study was applied in nature and, according to its approach, quantitative, with a preexperimental design. The population consisted of 100 infants; however, a non-probabilistic convenience sampling method was used, selecting a sample of 52 infants. Finally, the results showed that the system achieved an 86% accuracy rate in facial landmark detection. Regarding comprehensive emotional classification, the system obtained an overall accuracy of 33.33% and an F1-Score of 32.54%. These metrics, while lower than laboratory models, demonstrate the feasibility of implementing lightweight architectures for supporting the interpretation of infant needs on low-cost hardware.

Instituciones: Universidad Privada del Norte