Abstract: This work presents a lightweight fusion approach integrating camera and LiDAR data aimed at enhancing scene understanding of automated driving in rural terrains. The proposed method leverages the complementary strengths of both modalities while addressing the challenges of computational efficiency and real-time performance. By fusing the high-resolution visual information from cameras with the accurate depth perception provided by LiDAR point cloud, the approach achieves robust scene understanding of a 3D object in its environment. Furthermore, the method introduces a lightweight fusion architecture designed to minimize computational overhead without sacrificing accuracy. Experimental results demonstrate the effectiveness and efficiency of the proposed method, highlighting its potential for scene understanding of complex rural environments, enabling safe and efficient navigation on rural roads.