A novel microwave sensor and deep learning approach for rapid and non-destructive evaluation of egg freshness

Abstract

This study introduces a novel planar microwave sensing framework, coupled with deep learning, for rapid and non-invasive evaluation of egg freshness. The sensor, engineered with an Omega split-ring resonator (OSRR) and operating across 1.5–6.5 GHz, captures subtle dielectric variations reflective of internal quality changes. A total of 20 eggs were tracked under controlled conditions across four sensor placements, with the blunt end yielding superior spectral stability. Spectral and physical features were fused and processed via a one-dimensional convolutional neural network, achieving a coefficient of determination up to 0.951 and a root mean square error of 0.136. Clustering and visualization techniques further validated the discriminative power of the system. By uniting high-frequency microwave sensing with intelligent feature learning, this work establishes a robust, real-time pathway for industrial-scale, non-destructive freshness monitoring—offering a transformative solution for modern food quality assurance.