Panoster: End-to-end Panoptic Segmentation of LiDAR Point Clouds

TL;DR

Panoster is a fast, end-to-end LiDAR point cloud panoptic segmentation method that uses learning-based clustering, achieving state-of-the-art accuracy without post-processing.

Contribution

It introduces a simplified, proposal-free framework with a learning-based clustering approach for LiDAR panoptic segmentation, outperforming prior methods.

Findings

Achieved state-of-the-art results on SemanticKITTI benchmark.

Outperformed previous methods in accuracy while maintaining speed.

Flexible integration with existing semantic architectures.

Abstract

Panoptic segmentation has recently unified semantic and instance segmentation, previously addressed separately, thus taking a step further towards creating more comprehensive and efficient perception systems. In this paper, we present Panoster, a novel proposal-free panoptic segmentation method for LiDAR point clouds. Unlike previous approaches relying on several steps to group pixels or points into objects, Panoster proposes a simplified framework incorporating a learning-based clustering solution to identify instances. At inference time, this acts as a class-agnostic segmentation, allowing Panoster to be fast, while outperforming prior methods in terms of accuracy. Without any post-processing, Panoster reached state-of-the-art results among published approaches on the challenging SemanticKITTI benchmark, and further increased its lead by exploiting heuristic techniques. Additionally, we showcase how our method can be flexibly and effectively applied on diverse existing semantic architectures to deliver panoptic predictions.