4D printed deformation labels with machine learning for monitoring and preservation of respiring climacteric fruits

Abstract 4D printed labels that change color and shape were developed to achieve the dual functions of quality assessment and maintenance of respiring climacteric fruits. The effect of addition of essential oil emulsion and different geometric structures on deformation as well as the corresponding mechanisms were explored. Cast, 3D printed, and 4D printed labels were compared based on their responses to fruit quality and compatibility with machine learning. The addition of emulsion significantly affected the degree of deformation by altering the printing fidelity, hydrophilicity, and flexibility of the network structure. Geometric designs (Including printing layers, filament intersection angles, and infill ratios) changed both the direction and degree of deformation. Unlike cast and 3D printed labels, 4D printed labels simultaneously changed color and shape in response to variations in humidity and carbon dioxide levels in the package, enabling more accurate visual monitoring the turning points of fruit quality. The MobileNet model achieved recognition accuracy of 97% for 4D printed labels, which played an active role in achieving intelligent warnings. Additionally, deformation along with microstructure destruction positively impacted the controlled release of essential oils through a non-Fickian mechanism, resulting in a better preservation effect.