Abstract:
Maize is a staple food in Sub-Saharan Africa, and tillage is widely
used to boost its yield, though it affects soil and the environment both
positively and negatively. To support farmers and policymakers, a data-driven
approach using UAV technology was introduced.
This study was conducted for two seasons in a randomized complete
block design with four treatments (Harrowing only, Ploughing only,
Ploughing and Harrowing, and No-tillage). The results showed that No-tillage
had the lowest growth parameters, while Ploughing and Harrowing recorded
the highest in terms of LAI (1.50–1.75), stem diameter (20–22.5 mm), plant
height (165–175 cm), and yield (7.20–10.93 t/ha biomass, 4.619–5.67 t/ha
grain yield). Despite its lower yields, No-tillage showed the highest yield
improvement (+1.11 t/ha). UAVs imagery with Yolov8-small achieved high
germination rate detection (mAP50: 0.89–0.95) and accurate plant height
estimation (RMSE < 7 cm, R²: 0.98–0.99). For LAI estimation, UAV
technology coupled with Huber regression model achieved R² scores of 0.80–
0.94 and RMSE as low as 0.14, and coupled with Gradient Boosting Machines
reached R² of 0.87 and RMSE of 0.281 t/ha at the vegetative stage for Yield
prediction. Ploughing and Harrowing is recommended for short-term tillage,
while No-tillage is better for the long term. UAV imagery with machine
learning reliably monitors maize and predicts yield. Future research should
explore the long-term effects of No-tillage, UAV-based stem girth estimation,
and the cost-benefit of UAV adoption in small-scale farming.
Keywords: Tillage, UAV technology, maize, yield prediction, maize.
