A groundbreaking study leverages UAV hyperspectral imaging and machine learning to accurately assess nicotine levels in cigar leaves.
Recent advancements in aerial hyperspectral imaging, combined with machine learning, have revolutionized nicotine monitoring in cigar leaves. This cutting-edge approach enhances assessment accuracy while providing valuable insights for the tobacco industry, where chemical composition is critical to quality.
Led by Tian et al. at Sichuan Agricultural University, researchers sought to overcome the limitations of traditional manual quality checks, which often lack precision and efficiency. Their study, published on February 2, 2025, identifies strong correlations between nitrogen fertilizer use, moisture levels, and nicotine concentrations, underscoring the importance of timely and precise monitoring techniques.
The study was conducted from May to September 2022 at the university’s Modern Agricultural Research Base, where researchers used unmanned aerial vehicles (UAVs) equipped with hyperspectral cameras to capture leaf reflectance spectra from 15 different cigar leaf varieties under various nitrogen treatments.
Their findings revealed a direct correlation between nitrogen fertilizer application and nicotine levels in cigar leaves. “With the increase in the rate of application of nitrogen fertilizer, the nicotine content of cigar leaves increased,” the authors stated, highlighting the impact of agricultural practices on product quality.
To enhance the quality of hyperspectral image data collected by UAVs, the study employed preprocessing techniques such as multivariate scatter correction, standard normal transformation, and Savitzky-Golay convolution smoothing. Advanced machine learning algorithms, including Partial Least Squares Regression (PLSR) and Back Propagation neural networks, were then applied to develop predictive models capable of accurately estimating nicotine content.
The most effective model identified was the MSC-SNV-SG-CARS-BP model, which achieved a testing accuracy with R² values of approximately 0.797 and an RMSE of 0.078. “The MSC-SNV-SG-CARS-BP model has the best predictive accuracy on the nicotine content,” the authors noted, positioning it as a promising tool for future research and precision agriculture applications.
By utilizing remote sensing to analyze the spectral properties of cigar leaves, farmers and producers can assess crop quality swiftly and non-destructively, enabling more informed production and supply chain decisions. This approach offers extensive coverage at low operational costs while ensuring data consistency by reducing reliance on human factors.
The integration of hyperspectral imaging and machine learning has the potential to transform traditional tobacco cultivation, not only enhancing nicotine quality but also promoting sustainable and efficient agricultural practices. Researchers emphasize the need for continued advancements to refine these technologies and adapt them for different tobacco varieties and other crops.
Future studies will focus on optimizing UAV operational conditions to capture the highest-quality spectral data, considering variables such as flight altitude, lighting conditions, and noise reduction. Addressing these factors is crucial as agricultural practices evolve to meet market demands while prioritizing environmental sustainability.
This research highlights the synergy between technology and agricultural science, underscoring the growing adoption of innovative techniques to improve product quality. The researchers advocate for broader applications of hyperspectral sensing across agriculture, reinforcing the role of technology in enhancing yield, efficiency, and environmental responsibility.
Sources: https://www.nature.com/articles/s41598-025-88091-4
Crop monitoring
