Fast-BEV++: Fast by Algorithm, Deployable by Design

TL;DR

Fast-BEV++ introduces a hardware-efficient, real-time BEV perception framework that significantly improves speed and accuracy for autonomous driving, enabling deployment across diverse automotive platforms.

Contribution

It proposes a novel, hardware-oriented view transformation module and an integrated depth module, achieving over 3x speedup and state-of-the-art accuracy on nuScenes.

Findings

Achieves 0.488 NDS on nuScenes benchmark.

Delivers real-time inference at over 134 FPS.

Eliminates custom kernel dependencies for broader hardware compatibility.

Abstract

The advancement of vision-only Bird's-Eye-View (BEV) perception, a core paradigm for cost-effective autonomous driving, is hindered by the long-standing fundamental trade-off between perception accuracy and on-device deployment efficiency. In this work, we introduce Fast-BEV++, a BEV perception framework that resolves this tension through two fundamental design principles: Fast by Algorithm and Deployable by Design. By decomposing the core view transformation module into a hardware-oriented standard Index-Gather-Reshape pipeline, Fast-BEV++ eliminates dependencies on custom kernels while achieving no less than 3 times speedup over the Fast-BEV baseline across mainstream edge platforms. Empirically, Fast-BEV++ establishes a new state-of-the-art result of 0.488 NDS on the nuScenes 3D object detection benchmark, simultaneously delivering real-time inference at more than 134 FPS via our acceleration design. In particular, our integrated, learnable depth module yields consistent performance gains, maintaining the highest accuracy among comparable methods. Overall, this inherently decomposed architecture enables seamless real-time deployment across diverse production-grade automotive platforms, alleviating hardware limitations without compromising perception accuracy or inference efficiency.