RTB-YOLO: a lightweight UAV-based detection model for pine red turpentine beetle infestation

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The real-time and precise detection of Red Turpentine Beetle (RTB) infections is crucial for mitigating ecological and economic losses of pine forests. However, RTB infestation images from unmanned aerial vehicles always suffer from small target size as well as complex background interference, increasing the difficulty of high-precision detection under lightweight constraints. To address this challenge, RTB-You Only Look Once (RTB-YOLO), a lightweight model for UAV-based pine RTB infestation detection based on YOLO11n is proposed. In RTB-YOLO, a multi-scale partial feature aggregation module with global characteristics is designed to expand the effective receptive field and enhance the model’s feature representation capability. Furthermore, a spatial–channel fusion network is constructed as the neck architecture to facilitate efficient contextual information exchange while substantially reducing model complexity. Finally, a lightweight feature extraction head is developed to improve localization accuracy for critical regions while further reducing the computational overhead of the detection head. On the Pests and Diseases Tree dataset, RTB-YOLO achieves a mean Average Precision of 94.1% at IoU 0.5 and 67.3% at IoU 0.5:0.95, with an inference speed of 224.09 frames per second. Meanwhile, the number of model parameters, floating-point computation, and model weights are reduced by 36.05%, 12.7%, and 33.94%, respectively. Experimental results suggest that the model demonstrates excellent stability and robustness across diverse complex scenarios, providing reliable technical support for intelligent detection and monitoring of RTB infestation in forestry.