Surround-view Fisheye BEV-Perception for Valet Parking: Dataset, Baseline and Distortion-insensitive Multi-task Framework

TL;DR

This paper introduces a large-scale fisheye parking dataset and a real-time multi-task perception network that effectively handles fisheye distortion and environmental challenges in valet parking scenarios.

Contribution

The paper presents a new fisheye parking dataset (FPD) and a distortion-insensitive multi-task framework (FPNet) for improved surround-view perception in valet parking.

Findings

FPD dataset covers diverse real-world parking scenarios.

FPNet achieves real-time performance with enhanced distortion handling.

Experimental results show superior perception accuracy and generalizability.

Abstract

Surround-view fisheye perception under valet parking scenes is fundamental and crucial in autonomous driving. Environmental conditions in parking lots perform differently from the common public datasets, such as imperfect light and opacity, which substantially impacts on perception performance. Most existing networks based on public datasets may generalize suboptimal results on these valet parking scenes, also affected by the fisheye distortion. In this article, we introduce a new large-scale fisheye dataset called Fisheye Parking Dataset(FPD) to promote the research in dealing with diverse real-world surround-view parking cases. Notably, our compiled FPD exhibits excellent characteristics for different surround-view perception tasks. In addition, we also propose our real-time distortion-insensitive multi-task framework Fisheye Perception Network (FPNet), which improves the surround-view fisheye BEV perception by enhancing the fisheye distortion operation and multi-task lightweight designs. Extensive experiments validate the effectiveness of our approach and the dataset's exceptional generalizability.