Concerning the issues that existing infrared and visible image fusion algorithms inadequately extract detailed features from source images and fail to consider the relationship between the fusion network and the detection network, an improved infrared and visible image fusion algorithm based on RepVGG is proposed in this paper. Firstly, the network architecture of PIAfusion is used to replace the convolutional block of the image feature extraction part of the network architecture, and the feature reconstruction convolutional block part of the network architecture with the RepVGG convolutional block. Subsequently, the YOLOv5 detection network is employed to detect the fused images, with YOLOv5 being utilized to build detection losses. Then, the detection loss is leveraged to guide the training of the fusion network through backpropagation, ensuring that the fused images output by the fusion network can be more easily detected by the detection model. Finally, the detection results of the fusion images output by the fusion network in the YOLOv5 detection network are obtained. Compared with the existing fusion methods, the results show that the fusion image obtained by the proposed method has a good effect from the objective indicators and the detection results of YOLOv5.