An AI-driven quality assessment, pesticide residue investigation and chemical compound analysis system for clustering pepper products.

Abstract

This study investigated the colour values (L*, a*, b*, C* and H°), pesticide residue (PR) levels, and chemical compound compositions of red pepper flakes, red pepper powder, and flakes of isot, a distinctive regional type of chilli pepper. PR levels and chemical component compositions of the spices were determined. The L* values depicted that powdered peppers were generally the lightest in colour, followed by red pepper flakes, and the darkest were isot peppers. An AI (artificial intelligence) driven unsupervised learning approach, self-organising maps (SOM) was employed to cluster pepper products and the prompt detection of counterfeit by colour analysis. The SOM extracted the colour features of peppers represented in hexagonal lattices for product identification and different quality rates which make it possible to identify identical pepper products and show their differences from other pepper varieties. SOM is a potent AI tool for confirming the expected linkage between pepper qualities and forecasting unknown correlations and maps to enable quick, visual, and qualitative understanding of pepper attribute connections. The amount of PR in isot, red pepper flakes, and red powdered pepper was analysed using an LC-MS/MS device for 403 active pesticide substances. PR analysis revealed the presence of nicotine, chlorantraniliprole, tebuconazole, fluxapiroxad, pyraclostrobin, and azoxystrobin that were below the Maximum Residue Limit values of Turkish Food Codex and EU. Overall, GC-MS analysis showed that carboxylic acid compounds were the predominant compound group in all spice samples, followed by alkane compounds.