Abstract:

Given that traditional forest pest monitoring methods mainly rely on manual inspections and ground surveys，which have problems such as low efficiency，high cost，and limited coverage，this paper proposes a real-time monitoring method for forest pest disasters based on multispectral remote sensing and improved YOLOv8. Multispectral remote sensing images of forest pest disasters are acquired using a drone equipped with a multispectral camera， followed by smoothing and transformation processing to enhance image quality and reduce noise interference. By introducing attention mechanism to improve the conventional YOLOv8 model，and using the improved model to classify and recognize multispectral remote sensing images，real-time monitoring of forest pest disaster levels can be achieved. The case analysis results indicate that the research method can balance high real-time monitoring accuracy and speed of forest pest disasters，meeting the prevention and control needs of large-scale forest pest disasters.

Key words: multispectral remote sensing, improved YOLOv8, forest pest disasters, disaster monitoring, real time monitoring