Edge AI for Real-Time Facial Emotion Detection: A Raspberry Pi-Based Approach with CNN

Abstract

Forecasts and analysis of human behavior and emotions have received a lot of development in the area of artificial intelligence. There are some potentials that edge computing in the field of human behavior and emotion forecasting and analysis. In this research, an autonomous edge computing system utilizing Raspberry Pi for real-time emotion detection, which can detect seven basic emotions, namely, happiness, anger, sadness, surprise, neutral, fear, and disgust, has been developed. The system extracts several feature sets such as edges, shapes, and textures, as well as landmarks of the face and upper body. A convolutional neural network classifier is applied to these feature sets to classify emotions in real time. A classification model has been trained using a manually developed dataset through the Data Collector class. Real-time testing of the system and evaluation of the system based on several criteria such as F-measure, accuracy, precision, and sensitivity have been performed. It has been found that the proposed system has up to 98% accuracy in real-time emotion detection.



Received: 21 May 2026 | Revised: 1 September 2026 | Accepted: 15 September 2026



Conflicts of Interest

The authors declare that they have no conflicts of interest related to this work.



Data Availability Statement

Data are available from the corresponding author upon reasonable request.



Author Contribution Statement

Muneera Altayeb: Conceptualization, Methodology, Software, Resources, Writing – original draft, Supervision, Project administration. Khaled Abu Al Rub: Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing – original draft. Ahmad Nazzal: Validation, Formal analysis, Investigation, Data curation, Visualization. Areen Arabiat: Writing – review & editing, Visualization.