High-resolution crop type mapping in Ghana
Figure: Existing crop labels data captured from GoPro camera and UAV for 3 consecutive cropping seasons (2023, 2024 and 2025).
Timely and accurate data on the extent of croplands and the types of cultivated crops are essential to addressing the food insecurity challenges that many African countries face. This information supports the provision of extension services to smallholder farmers. The precise and timely identification of crop types is necessary for effective food security monitoring and planning. However, crop type mapping in small-scale farming systems in Ghana is challenging due to relatively small and heterogeneous farms, and limited crop labels. The challenge can be addressed by applying new methods for the rapid collection of crop labels and leveraging advancements in machine learning algorithms to process and classify big data. The proposed study will use high-resolution crop labels digitized from car-mounted GoPro cameras and unmanned aerial vehicle (UAV), to classify crop types from a combination of vegetation indices generated from the Sentinel-1, Sentinel-2, and PlanetScope sensors in Northern Ghana.
The main aim of the proposed study is to apply machine learning for crop type classification in a mixed farming system in Northern Ghana. The specific objectives are to: (1) digitize crop labels for major cereal and legume crops (maize, rice, soyabean and groundnut) from geotagged images captured from GoPro cameras and UAVs, (2) generate crop-type maps from crop labels and a synergistic combination of high-resolution satellite data from optical and active sensors, and (3) accuracy assessment of the crop-type classification, (4) external validation of crop type maps with data obtained from local government statistics and existing open-source databases.
Huang, X., Vrieling, A., Dou, Y., Li, X., and Nelson, A. (2025). Divergent crop mapping accuracies across different field types in smallholder farming regions. International Journal of Applied Earth Observation and Geoinformation 139, 104559.
Nakalembe, C., Zvonkov, I., Kerner, H., Frimpong, D. B., Mwangi, K., Kioko, J., . . . Loh, P. M. (2025). Helmets Labeling Crops: Kenya Crop Type Dataset Created via Helmet-Mounted Cameras and Deep Learning. Scientific Data, 12(1), 1496. doi:10.1038/s41597-025-05762-7
Paliyam, M., Nakalembe, C.L., Liu, K., Nyiawung, R. and Kerner, H.R., 2021. Street2Sat: A Machine Learning Pipeline for Generating Ground-truth Geo-referenced Labeled Datasets from Street-Level Images, 38th International Conference on Machine Learning, PMLR, pp. 139.