Evaluating UAV Multispectral Imagery, Machine Learning and Image Analysis Techniques for Mapping Taro and Sweet Potato in a Smallholder Cropland in Swayimane, South Africa
Bibliographic Data
| ID | 17819758 |
|---|---|
| Authors | Mishkah Abrahams (0000-0003-4068-2141, University of the Western Cape, corresponding author), Mbulisi Sibanda (0000-0002-4589-7099, University of the Western Cape), Timothy Dube (0000-0003-3456-8991, University of the Western Cape), James Magidi (0000-0001-5887-0904, Tshwane University of Technology), Richard Kunz (0000-0001-8372-6218, University of KwaZulu-Natal), Vimbayi G P Chimonyo (International Maize and Wheat Improvement Center), Alistair D Clulow (University of KwaZulu-Natal), Tafadzwanashe Mabhaudhi (0000-0002-9323-8127, London School of Hygiene & Tropical Medicine) |
| Year | 2026 |
| Volume | 13 |
| Issue | 1 |
| Publication date | 2026-01-01 |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Geo Geography and Environment (JOURNAL) |
| Journal identifiers | ISSN: 2054-4049 • E-ISSN: 2054-4049 |
| Publisher | Wiley (PUBLISHER • GB) |
| DOI | 10.1002/geo2.70078 |
| OpenAlex | W7154711995 |
| Language | EN |
| References cited | 68 |
Accurate mapping of neglected and underutilised crop species (NUS) in complex smallholder fields is critical for enhancing food security, yet it remains a significant challenge due to landscape heterogeneity, spectral similarities between crops and the spatial complexity of strip intercropping systems. This study evaluates the efficacy of unmanned aerial vehicle (UAV) multispectral imagery for mapping taro and sweet potato, addressing the limitations of traditional pixel‐based image analysis (PBIA), which often fails in fragmented agricultural landscapes. We quantitatively compare the performance of PBIA and object‐based image analysis (OBIA) frameworks using a Gradient Tree Boost (GTB) classifier and determine the optimal input dataset by testing raw spectral bands, vegetation indices (VIs) and their combination. The findings demonstrate the clear superiority of the OBIA–GTB approach: using the combined dataset, OBIA–GTB achieved a mean overall accuracy of 95% ± 3.66%, kappa = 0.9419. In contrast, PBIA‐GTB yielded 83% ± 7.10% overall accuracy, with kappa = 0.8016. This study validates a robust, scalable workflow for NUS mapping, concluding that the integration of high‐resolution UAV data, OBIA and GTB provides a precise solution for agricultural monitoring in heterogeneous, intercropped smallholder systems
Agriculture · Contextual image classification · Decision tree · Intercropping · Multispectral image · Precision agriculture · Random forest · Workflow · Land Use and Ecosystem Services · Remote Sensing in Agriculture · Remote-Sensing Image Classification
Geographic Object-Based Image Analysis – Towards a new paradigm
Making better use of accuracy data in land change studies
Gradient boosting machines, a tutorial
Object based image analysis for remote sensing
The Measurement of Observer Agreement for Categorical Data
On the emergence of geospatial cloud-based platforms for disaster risk management
Future Smart Food
| Highly cited | No |
|---|