BEVFormer v2: Adapting Modern Image Backbones to Bird's-Eye-View Recognition via Perspective Supervision

TL;DR

This paper introduces a novel bird's-eye-view detector with perspective supervision that improves convergence and compatibility with modern image backbones, achieving state-of-the-art results on nuScenes.

Contribution

It proposes a two-stage BEV detector with perspective supervision, enabling better integration of modern image backbones and faster, more effective training.

Findings

Achieves new state-of-the-art on nuScenes dataset.

Compatible with a wide range of image backbones.

Faster convergence and improved performance.

Abstract

We present a novel bird's-eye-view (BEV) detector with perspective supervision, which converges faster and better suits modern image backbones. Existing state-of-the-art BEV detectors are often tied to certain depth pre-trained backbones like VoVNet, hindering the synergy between booming image backbones and BEV detectors. To address this limitation, we prioritize easing the optimization of BEV detectors by introducing perspective space supervision. To this end, we propose a two-stage BEV detector, where proposals from the perspective head are fed into the bird's-eye-view head for final predictions. To evaluate the effectiveness of our model, we conduct extensive ablation studies focusing on the form of supervision and the generality of the proposed detector. The proposed method is verified with a wide spectrum of traditional and modern image backbones and achieves new SoTA results on the large-scale nuScenes dataset. The code shall be released soon.