Integrating fluorescence enhancement and adsorption enrichment: Nanofibrous mat sensor with deep learning-assisted data analysis for real-time monitoring of shrimp freshness.

Herein, a high-performance fluorescent nanofibrous mat (NFM) integrating β-cyclodextrin (β-CD) and adamantane-modified probes was developed for real-time visual monitoring of food freshness. As a functional additive, β-CD forms inclusion complexes with fluorescent probes to enhance fluorescence intensity. Meanwhile, it efficiently enriches biogenic amines (BAs) and further amplifies the sensing response. The sensor showcases rapid responsiveness (<30 s), ultra-low detection limits (78.7-253.2 ppb for five BAs), excellent selectivity and reusability, accompanied by a distinguishable color transition from red to green. To further improve the recognition ability, a deep learning platform based on the DenseNet-121 convolutional neural network was established, which can classify shrimp freshness into fresh, less-fresh, and spoiled with an accuracy exceeding 99%. For practical application, a user-friendly smartphone mini-program integrating CNN analysis was further developed. Through demonstrating the enhancement of solid-phase sensing via rational substrate engineering, this integrated system offers a promising paradigm for real-time food quality monitoring.