单帧实时重建道路表面，精度达1.73厘米，适合自动驾驶场景。
BEV-GS: Feed-forward Gaussian Splatting in Bird's-Eye-View for Road Reconstruction
- 基于俯视图设计几何与纹理分离网络，单帧直接预测参数。
- 道路高程误差仅1.73厘米，新视角合成PSNR达28.36 dB。
- 推理速度达26帧/秒，适合车载实时系统部署。
道路是车轮或机器人足部唯一接触介质，其重建对无人驾驶车辆和移动机器人至关重要。近期神经辐射场（NeRF）与高斯点阵（GS）在场景重建中取得显著成果，但通常依赖多视角图像输入且需长时间优化。本文提出BEV-GS，一种基于前馈高斯点阵的实时单帧道路表面重建方法。BEV-GS包含预测模块与渲染模块：预测模块采用俯视图范式，分别设计几何与纹理网络，直接从单帧估计参数，避免逐场景优化；渲染模块采用网格高斯表示道路表面并实现新视角合成，更契合道路特征。该方法在真实数据集RSRD上达到当前最优性能，道路高程误差降低至1.73厘米，新视角合成的PSNR达28.36 dB。预测与渲染帧率分别为26和2061，支持高精度与实时应用。代码将开源于：https://github.com/cat-wwh/BEV-GS
原文摘要 · Abstract (English)
Road surface is the sole contact medium for wheels or robot feet. Reconstructing road surface is crucial for unmanned vehicles and mobile robots. Recent studies on Neural Radiance Fields (NeRF) and Gaussian Splatting (GS) have achieved remarkable results in scene reconstruction. However, they typically rely on multi-view image inputs and require prolonged optimization times. In this paper, we propose BEV-GS, a real-time single-frame road surface reconstruction method based on feed-forward Gaussian splatting. BEV-GS consists of a prediction module and a rendering module. The prediction module introduces separate geometry and texture networks following Bird's-Eye-View paradigm. Geometric and texture parameters are directly estimated from a single frame, avoiding per-scene optimization. In the rendering module, we utilize grid Gaussian for road surface representation and novel view synthesis, which better aligns with road surface characteristics. Our method achieves state-of-the-art performance on the real-world dataset RSRD. The road elevation error reduces to 1.73 cm, and the PSNR of novel view synthesis reaches 28.36 dB. The prediction and rendering FPS is 26, and 2061, respectively, enabling high-accuracy and real-time applications. The code will be available at: \href{https://github.com/cat-wwh/BEV-GS}{\texttt{https://github.com/cat-wwh/BEV-GS}}
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。