Monitoring green tea fixation quality by intelligent sensors: comparison of image and spectral information

Abstract

BACKGROUND Intelligent monitoring of fixation quality is a prerequisite for automated green tea processing. To meet the requirements of intelligent monitoring of fixation quality in large‐scale production, fast and non‐destructive detection means are urgently needed. Here, smartphone‐coupled micro near‐infrared spectroscopy and a self‐built computer vision system were used to perform rapid detection of the fixation quality in green tea processing lines. RESULTS Spectral and image information from green tea samples with different fixation degrees were collected at‐line by two intelligent monitoring sensors. Competitive adaptive reweighted sampling and correlation analysis were employed to select feature variables from spectral and color information as the target data for modeling, respectively. The developed least squares support vector machine (LS‐SVM) model by spectral information and the LS‐SVM model by image information achieved the best discriminations of sample fixation degree, with both prediction set accuracies of 100%. Compared to the spectral information, the image information‐based support vector regression model performed better in moisture prediction, with a correlation coefficient of prediction of 0.9884 and residual predictive deviation of 6.46. CONCLUSION The present study provided a rapid and low‐cost means of monitoring fixation quality, and also provided theoretical support and technical guidance for the automation of the green tea fixation process. © 2022 Society of Chemical Industry.

T‐YOLO: a lightweight and efficient detection model for nutrient buds in complex tea‐plantation environments

“…The proportion of nutrient buds in tea raw materials is a crucial indicator of tea-leaf quality. [1][2][3] Early detection of nutrient buds is important so that tea gardens can monitor tea-tree growth and predict premium tea yield. 4 However, the identification of tea nutrient buds is still in the manual stage; monitoring large areas is time consuming and laborious, and climate and other factors limit its effectiveness.…”

“…The tea nutrient buds are the most economically valuable part of the tea plant. The proportion of nutrient buds in tea raw materials is a crucial indicator of tea‐leaf quality 1‐3 . Early detection of nutrient buds is important so that tea gardens can monitor tea‐tree growth and predict premium tea yield 4 .…”