A Systematic Literature Review for the Development of an Object Detection System for Monitoring Underground Crop (Garlic) Growth Using GPR

Abstract

This systematic literature review investigates the development of a Ground-Penetrating Radar (GPR)-based object detection system tailored for under-ground garlic crop monitoring. While garlic-specific GPR applications re-main limited, studies on structurally similar root crops such as potatoes and carrots provide a valuable reference framework. Using a PRISMA-guided methodology, 16 relevant studies were analysed and synthesized, highlighting advancements in GPR signal processing, object reconstruction, and machine learning integration. Results show that mid- frequency GPR (500–800 MHz), especially when paired with deep learning models such as 3D Convolutional Neural Networks (CNNs), offers high accuracy in detecting root structures. Key challenges such as signal attenuation in clay-rich and tropical soils are addressed through electromagnetic induction (EMI) hybridization and antenna optimization. A comparative matrix summarizes the most relevant findings, and actionable recommendations are proposed to guide future research. These include the development of garlic-specific datasets, localized field testing, and AI- enhanced signal classification. GPR, when effectively configured and paired with machine learning, presents a viable solution for real-time, non-invasive garlic crop monitoring in tropical agriculture.

Plant phenotyping relevance

地下作物の成長・根構造をGPRで検出する手法の開発に焦点を当てた系統的レビューであり、植物形態の非破壊取得・抽出方法が中心です。

Code and data availability

This is a systematic literature review of GPR-based underground crop monitoring. It presents no original phenotype datasets, GPR imagery, sensor data, analysis code, or trained models of its own; the authors explicitly call for the future creation of garlic-specific open-access datasets. The only URL (TuberScan) is a B

No evidence-backed public reproduction asset is currently recorded.