The outbreak of Pine Shoot Beetle (PSB, Tomicus spp.) posed a significant threat to the health of Yunnan pine forests, necessitating the development of an efficient and accurate remote sensing monitoring method. The integration of unmanned aerial vehicle (UAV) imagery and deep learning algorithms shows great potential for monitoring forest-damaged trees. Previous studies have utilized various deep learning semantic segmentation models for identifying damaged trees in forested areas; however, these approaches were constrained by limited accuracy and misclassification issues, particularly in complex forest backgrounds. This study evaluated the performance of five semantic segmentation models in identifying PSB-damaged trees (UNet, UNet++, PAN, DeepLabV3+ and FPN). Experimental results showed that the FPN model outperformed the others in terms of segmentation precision (0.8341), F1 score (0.8352), IoU (0.7239), mIoU (0.7185) and validation accuracy (0.9687). Under the pure Yunnan pine background, the FPN model demonstrated the best segmentation performance, followed by mixed grassland-Yunnan pine backgrounds. Its performance was the poorest in mixed bare soil-Yunnan pine background. Notably, even under this challenging background, FPN still effectively identified diseased trees, with only a 1.7% reduction in precision compared to the pure Yunnan pine background (0.9892). The proposed method in this study contributed to the rapid and accurate monitoring of PSB-damaged trees, providing valuable technical support for the prevention and management of PSB.

The pine shoot beetle (PSB, Tomicus spp.) is a destructive stem-boring pest threatening Yunnan pine ( Pinus yunnanensis ), a dominant conifer species in southwestern China. Early-stage infestation is particularly difficult to detect because only 11%−20% of shoots exhibit visible discoloration during the initial attack phase, limiting the effectiveness of conventional field surveys and single-source remote sensing approaches. Reliable early detection is therefore essential for timely intervention and outbreak prevention. This study assessed the potential of integrating unmanned aerial vehicle (UAV)-based multispectral and thermal infrared (TIR) imagery to improve early PSB detection at the individual-tree level. A total of 150 trees representing four damage levels (30 healthy, 50 lightly damaged, 40 moderately damaged, and 30 severely damaged trees) were identified through field-based visual interpretation. Twenty spectral and thermal metrics were extracted from UAV imagery, and commonly used machine learning approaches were applied under three feature scenarios: multispectral only, TIR only, and fused multispectral–TIR features, with 10-fold cross-validation. Most spectral and thermal variables differed significantly among infestation stages ( p < 0.05). The integration of multispectral and thermal features improved classification performance compared with single-source data. The optimal model (Random Forest) yielded an overall accuracy (OA) of 0.82 and a Kappa coefficient of 0.75. For lightly infested trees, Recall and Precision were 0.72 and 0.74, respectively. Vegetation indices (e.g., GRVI, NDVI, and RVI) were identified as primary predictors, while thermal variability metrics provided complementary information associated with early physiological stress. By enabling detection prior to widespread canopy damage, the proposed multisensor UAV framework supports timely forest health monitoring and provides a tool for integrated pest management under changing environmental conditions.