Grassland fires are characterized by their suddenness and difficulty in timely detection, which often bring great harm. Effective detection of grassland fire by UAV (unmanned aerial vehicle) is a new direction for future development. To address the problem of detecting and locating grassland fires by UAVs, we optimize the YOLOv8s model, improve its ability to identify grassland smoke and fire spots, guide UAVs to track possible targets, and improve the positioning accuracy of grassland fires by combining RTK (Real-Time Kinematic) information. After several field simulation tests and validation set tests, the results show that the optimized YOLOv8s model has high recognition ability for grassland smoke and fire spots, and the probability of the UAV to achieve high-precision fire spot localization can reach 90.82%.
Grassland Fire Detection and Localization of UAV Based on Optimized YOLOv8s
2025-05-16
6024719 byte
Conference paper
Electronic Resource
English
Vehicle detection research based on improved YOLOv8s
SPIE | 2025
|Deep Learning Network Model for Road Pavement Damage Detection Based on YOLOv8s-GES
Transportation Research Record | 2025
|