Intelligent storage year identification of Anhua dark tea via carbon quantum dots-based colorimetric sensor array and computer vision

Aging significantly influences Anhua dark tea's quality and value, but rapid year identification remains challenging. This study developed a colorimetric sensor array (CSA) combined with deep learning for the rapid identification of storage years of Anhua dark tea. The CSA uses five carbon quantum dots (CQDs) that interact specifically with tea components (e.g., amino acids, polyphenols), producing color changes with storage time; these changes can be captured by smartphones under UV illumination. The resulting images were directly analyzed using deep learning models (ResNet18, VGG16, DenseNet121, MobileNetV3-Small, and EfficientNet B0). Among them, ResNet18 achieved the highest accuracy (92.4%), outperforming traditional RGB-based models such as PLS-DA (57.6%) and PCA-LDA (63.6%). In addition, a user-friendly graphical user interface was created for easy, non-expert operation. This work presents a novel, cost-effective, and highly accessible detection platform for Anhua dark tea authentication, offering broad potential for real-world applications in food quality control and authenticity assessment.