Proposal-free Lidar Panoptic Segmentation with Pillar-level Affinity

TL;DR

This paper introduces a proposal-free lidar panoptic segmentation method that uses pillar-level affinity and local clustering to improve instance segmentation accuracy without requiring object detection annotations.

Contribution

It presents a novel proposal-free architecture combining semantic segmentation and affinity learning with a local clustering algorithm for lidar panoptic segmentation.

Findings

Outperforms previous proposal-free methods on nuScenes dataset.

Achieves comparable results to proposal-based methods without extra detection annotations.

Simplifies lidar panoptic segmentation with a pillar-based approach.

Abstract

We propose a simple yet effective proposal-free architecture for lidar panoptic segmentation. We jointly optimize both semantic segmentation and class-agnostic instance classification in a single network using a pillar-based bird's-eye view representation. The instance classification head learns pairwise affinity between pillars to determine whether the pillars belong to the same instance or not. We further propose a local clustering algorithm to propagate instance ids by merging semantic segmentation and affinity predictions. Our experiments on nuScenes dataset show that our approach outperforms previous proposal-free methods and is comparable to proposal-based methods which requires extra annotation from object detection.