Low-cost and dense fog-adapted monitoring for fermentation quality evaluation of black tea

Fermentation is a key step in black tea processing, which promotes the flavor formation of black tea. However, the in-situ and non-destructive quality monitoring within fermentation chamber, a dense-fog environment, is still lacking. Here, a novel indicator of fermentation degree (FD) was proposed, which is derived from epigallocatechin gallate (EGCG) content. Our results supported the feasibility of FD in characterizing fermentation quality across two sample batches, with an FD value at around 90 % suggesting satisfactory appearance availability for black tea. Additionally, a computer vision system was self-developed for real-time image acquisition within the high-humidity fermentation chamber. The deep learning-based Cycle-Dehaze achieved excellent defogging performance in varied fog concentrations, outperforming traditional dark channel prior. Color variables in the defogged images combined with a regression model achieved accurate prediction and visualization of FD, with residual predictive deviation (RPD) reaching 4.11. Another batch of validation samples further confirmed the model's reliability, achieving an RPD of 3.88. This study thus provides a low-cost solution for in-situ quality monitoring of black tea, benefiting optimal tea flavor.