Display omitted •The droplet characteristics of UAV based spraying system were evaluated.•ImageJ produced most consistent droplet characteristics among tested methods.•RSM techniques optimized the operating parameters of UAV-based spraying system.•GWO-ANN was the best neural networks architecture to predict droplet deposition. Modern agriculture relies on pesticides to increase crop yields, but these chemicals are also hazardous. Although conventional sprayers were designed for effective pest management, they nonetheless pollute the environment and endanger operators' health. Unmanned aerial vehicle (UAV)-based sprayers overcome the aforesaid problem and can precisely target the areas that need treatment and difficult to reach for human operators. This study evaluated a UAV-based spraying system in a cotton field, employing imaging techniques such as Laser Droplet Analyzer, Deposit Scan, ImageJ, and Drop leaf. Furthermore, the system was optimized using response surface methodology, and deposition predictive analysis was conducted using a hybrid GWO-ANN approach. The volume median diameter, number median diameter, relative span, and uniformity coefficient were in the range of 95–248 µm, 65–174 µm, 0.8–1.7 %, and 1.3–1.7 %, respectively. Optimizing the working speed (3.3 m/s), working height (1.0 m), and discharge rate (2.0 L/min) resulted in a droplet density of 50.3 droplets/cm2, deposition of 0.20 µL/cm2, and coverage of 9.27 %. The GWO-ANN prediction model yielded R2, RMSE, and MAE values of 0.878, 0.01729, and 0.01368, respectively. Optimizing operational parameters through multiple measurement techniques enhance flexibility and effectiveness of UAV-based spraying system, facilitating wider deployment in remote agricultural areas for agrochemical applications.