Year: 2025
Type: preprint
Abstract: Tospovirus infections in capsicum crops pose a serious threat to agricultural productivity, as infected seeds often appear visually indistinguishable from healthy ones.While conventional detection techniques such as DAS-ELISA offer high accuracy, they are time-consuming, labour-intensive, and require skilled personnel.This study proposes a non-invasive, non-destructive optical screening method for the rapid detection of Tospovirus-infected capsicum seeds.Optical sensor system integrates a custom-designed hardware setup comprising photodiodes and colour sensors illuminated by white and infrared light emitting diodes to capture reflectance data across red, green, blue, and infrared spectral bands.These reflectance signatures were processed using a Random Forest classification algorithm to distinguish infected seeds from healthy ones.Experimental results showed that the green spectral band delivered the highest classification performance, with an accuracy of 0.95, precision of 0.9667, recall of 0.9355, and F1 score of 0.9508.Red band also performed well, achieving 0.9 accuracy and an F1 score of 0.9062.These findings demonstrate the potential of the proposed optical screening approach as a rapid, accurate, and scalable method for early-stage detection of Tospovirus in seeds, offering a viable alternative to traditional diagnostics.
Authors: Shikha Sharma, Mansi Arora, Sushmita Mena, Saurav Bharadwaj
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: green
OpenAlex ID: W4414143415