Improved Single Camera BEV Perception Using Multi-Camera Training

TL;DR

This paper introduces a training method for BEV perception that enhances single-camera performance by leveraging multi-camera training techniques, including masking, cyclic learning rates, and reconstruction loss, to reduce hallucinations and improve map quality.

Contribution

The paper presents a novel training approach that improves single-camera BEV perception by transferring knowledge from multi-camera data using specific training strategies.

Findings

Outperforms single-camera models trained without multi-camera data

Reduces hallucination in BEV maps compared to baseline models

Achieves better BEV map quality with fewer cameras during inference

Abstract

Bird's Eye View (BEV) map prediction is essential for downstream autonomous driving tasks like trajectory prediction. In the past, this was accomplished through the use of a sophisticated sensor configuration that captured a surround view from multiple cameras. However, in large-scale production, cost efficiency is an optimization goal, so that using fewer cameras becomes more relevant. But the consequence of fewer input images correlates with a performance drop. This raises the problem of developing a BEV perception model that provides a sufficient performance on a low-cost sensor setup. Although, primarily relevant for inference time on production cars, this cost restriction is less problematic on a test vehicle during training. Therefore, the objective of our approach is to reduce the aforementioned performance drop as much as possible using a modern multi-camera surround view model reduced for single-camera inference. The approach includes three features, a modern masking technique, a cyclic Learning Rate (LR) schedule, and a feature reconstruction loss for supervising the transition from six-camera inputs to one-camera input during training. Our method outperforms versions trained strictly with one camera or strictly with six-camera surround view for single-camera inference resulting in reduced hallucination and better quality of the BEV map.