An intelligent fruit freshness monitoring system using hydrophobic indicator labels based on methylcellulose, k-carrageenan, and sodium tripolyphosphate, combined with deep learning.

As the demand for food quality and safety continues to rise, pH-responsive intelligent packaging technologies have found widespread application in the monitoring of food freshness. This study introduces a methylcellulose (MC)-based indicator label designed for real-time monitoring of fruit freshness. To enhance the label's structural integrity, mechanical properties, and hydrophobicity, k-carrageenan (KC) and sodium tripolyphosphate (STPP) were incorporated as crosslinkers to form a stable three-dimensional network structure. This modification led to a tensile strength of 23.33 MPa, an elongation at break of 113.4 %, and significant reductions in water content, solubility, and water vapor permeability, with the water contact angle increasing to 76.84°. To address the challenge in existing research where the ratio of mixed indicators requires time-consuming experimentation, the present study employs computer simulation techniques to model the color changes under varying chemical compositions and ratios, significantly reducing both experimental time and costs. Furthermore, an intelligent recognition method is proposed, involving the design of a label area cropping algorithm (ALC) integrated with a lightweight convolutional neural network (CNN), which effectively minimizes background interference and improves the accuracy of freshness detection. In experiments assessing the freshness of mango, kiwi, and grape, the method achieved maximum accuracies of 96.4 %, 96.3 %, and 96.3 %, respectively. Based on this approach, a mobile application has been developed for real-time monitoring of fruit freshness.