Camouflage object detection under low-light environment is one of the difficult problems in the field of uncovering camouflage, which often leads to degradation of the performance of conventional object detection algorithms if the light conditions are poor at this time. To address this problem, this paper proposes a camouflage object intelligent detection algorithm based on visible-infrared feature level fusion under low light. First, this paper improves the visible channel image quality through de-fogging illumination enhancement, which facilitates RGB feature information extraction; then, a feature-level fusion network guided by the object detection task is constructed, which utilizes residual connectivity to achieve multi-dimensional information extraction and stacking, and improves the efficiency of the information fusion utilization; finally, the camouflage object detection is improved through the optimization of the loss function, and the synthesis of the spatial-channel attention mechanism Effect. In this paper, an optical-infrared camouflage object dataset under low-light environment is constructed, and the proposed method is validated with real data, and the mAP@0.5 on this dataset is 92.3% and the accuracy P is 90.91%, which reflects the advantage of this paper"s algorithm in detecting camouflage object under low-light conditions. The method proposed in this paper greatly improves the low-light camouflage object detection effect, which can be applied to camouflage object detection in complex battlefield environments, and is of great practical significance for improving battlefield situational awareness.