| dc.description.abstract | This work explores the potential of neglected and underutilized crop species (NUS) in addressing agricultural, food, and nutrition security challenges exacerbated by climate change, particularly in Southern Africa. Mainstream crops like maize are adversely affected by climate variability, leading to increased insecurities. Despite the importance of NUS, limited research attention and market preference hinder their development. Additionally, there is a lack of criteria for determining their spatial extent in smallholder croplands, complicated by field fragmentation and intercropping. To overcome these challenges, this study employs unmanned aerial vehicles (UAVs) and high-throughput phenotyping technologies for accurate mapping of NUS, specifically sweet potato and taro, in smallholder farms in the Kwazulu-Natal Province, South Africa. Three specific objectives guide the study. These were (1) to conduct a systematic review of literature on the mapping the spatial distribution and health of NUS crops in sub-Saharan Africa, (2) to evaluate the performance of three robust classifiers in mapping the spatial distribution of NUS crops based on multispectral UAV data and, (3) to assess the performance of object based image analysis (OBIA) and pixel based analysis (PBIA) techniques combined with GTB classifier in mapping and delineating the spatial distribution of NUS crops. Review of literature revealed a lack of studies in the Global South, highlighting the potential of machine learning algorithms with optimal near-infrared and red-edge vegetation indices in mapping NUS. Despite slow progress due to high costs and regulations, the review findings suggested that integrating machine learning techniques with UAV-acquired data is crucial for efficient monitoring of NUS crops in small-scale agricultural areas. This will provide essential information for enhancing the efficiency of food production in small-scale agricultural areas located in the Global South. | en_US |
