I-TAINTED: Identification of Turmeric Adulteration Using the CavIty PerturbatioN Technique and Technology OptimizED Machine Learning

Abstract

In this work we propose a method to detect turmeric adulteration using the Cavity Perturbation Technique (CPT) at 2.4GHz. Two different adulterants are examined (egg-yellow color and starch). We show that when a single adulterant is added, the resonant frequency and unloaded quality factor values follow clear trends as a function of added contaminant. Unfortunately, when the turmeric is adulterated with different concentrations of two adulterants (e.g., a 50% color/50% starch) CPT does not lead to good results. To address this, we also present an automated machine learning flow that dramatically enhances the adulteration detection. The proposed flow has the additional uniqueness that it optimizes the predictive model based on the selected target hardware platform doing technology independent as well as technology dependent model optimizations. Experimental results show that our predictive model can be optimized based on the accuracy required for different hardware platforms. In particular we target a microcontroller and a dedicated hardware implementation.

Keywords

• Predictive model
• sensor design
• tumeric contamination