The menace of saffron adulteration: Low-cost rapid identification of fake look-alike saffron using Foldscope and machine learning technology

• DOI: 10.3389/fpls.2022.945291
• Publication Date: 2022-08-12T06:40:48Z
• Authors: Amjad M. Husaini; Syed Anam Ul Haq; Asma Shabir; Amir B. Wani; Muneer A. Dedmari
• Keywords: Identification; Robustness

Saffron authenticity is important for the saffron industry, consumers, food and regulatory agencies. Herein we describe a combo of two novel methods to distinguish genuine from fake in user-friendly manner without sophisticated instruments. A smartphone coupled with Foldscope was used visualize characteristic features “genuine” “fake.” Furthermore, destaining staining agents were study patterns. Toluidine blue pattern distinct easier use as it stained papillae margins deep purple, while its stain lighter yellowish green toward central axis. Further automate process, tested compared different machine learning-based classification approaches performing automated into or fake. We demonstrated that models are efficient learning morphological classifying samples either genuine, making much end-users. This approach performed better than conventional (random forest SVM), model achieved an accuracy 99.5% precision 99.3% on test dataset. The process has increased robustness reliability authenticating samples. first describes customer-centric frugal science-based creating app detect adulteration. survey conducted assess adulteration quality. It revealed only 40% belonged ISO Category I, average percentage remaining 36.25%. After discarding adulterants crude samples, their quality parameters improved significantly, elevating these category III II. Conversely, also means Categories II more prone favored by fraudsters.