Research on YOLO-DBO Object Detection Method Based on 1064-nm Laser Irradiation Locking

In this study, a YOLO-DBO cooperative perception framework integrating 1064 nm active laser labeling was proposed to address the difficulty of recognizing similar targets in complex scenes. The framework is based on constructing a target simulation dataset under laser tagging, dynamically fusing cross-scale features through the dimensional awareness selective integration (DASI) module, constructing the global spatial correlation between the laser spot and the background using the multiple self-attention (MHSA) mechanism, and enhancing the feature representation of the locally occluded target through the full-dimensional dynamic convolution (ODConv). Experiments demonstrate that this active-passive collaborative architecture reaches 96.1% in [email protected], which is 1.3% and 3.7% higher than YOLOv9 and RT-DETR, respectively, confirming the significant advantage of laser marker-based multimodal feature fusion in recognizing similar targets in complex environments.