Rapid detection of raw meat freshness using deep learning and colorimetric/fluorescent array

Effective monitoring of raw meat freshness is essential for food safety and reducing waste. This study presents a rapid freshness detection system combining a colorimetric fluorescent array (CFA) with deep learning (DL). The CFA, made by loading alizarin and fluorescein isothiocyanate (FITC) onto a polyvinylidene fluoride (PVDF) membrane, provides a sensitive response to amines produced during meat spoilage. A dataset of 8170 images was used to train convolutional neural networks (CNN) and Vision Transformer models. The CNN-based ResNet-152 model achieved the highest prediction accuracy, with 95.82% under natural light and 97.88% under ultraviolet light. The system can assess meat freshness from a single image in 130 ms. This CFA-DL approach offers a fast, non-destructive, and accurate method for real-time monitoring of meat freshness, potentially enhancing food safety and reducing waste without the need for specialized personnel or equipment.