LESS: Label-Efficient Semantic Segmentation for LiDAR Point Clouds

10/14/2022
by   Minghua Liu, et al.
2

Semantic segmentation of LiDAR point clouds is an important task in autonomous driving. However, training deep models via conventional supervised methods requires large datasets which are costly to label. It is critical to have label-efficient segmentation approaches to scale up the model to new operational domains or to improve performance on rare cases. While most prior works focus on indoor scenes, we are one of the first to propose a label-efficient semantic segmentation pipeline for outdoor scenes with LiDAR point clouds. Our method co-designs an efficient labeling process with semi/weakly supervised learning and is applicable to nearly any 3D semantic segmentation backbones. Specifically, we leverage geometry patterns in outdoor scenes to have a heuristic pre-segmentation to reduce the manual labeling and jointly design the learning targets with the labeling process. In the learning step, we leverage prototype learning to get more descriptive point embeddings and use multi-scan distillation to exploit richer semantics from temporally aggregated point clouds to boost the performance of single-scan models. Evaluated on the SemanticKITTI and the nuScenes datasets, we show that our proposed method outperforms existing label-efficient methods. With extremely limited human annotations (e.g., 0.1 even highly competitive compared to the fully supervised counterpart with 100 labels.

READ FULL TEXT
research
03/16/2022

Scribble-Supervised LiDAR Semantic Segmentation

Densely annotating LiDAR point clouds remains too expensive and time-con...
research
07/15/2022

LapSeg3D: Weakly Supervised Semantic Segmentation of Point Clouds Representing Laparoscopic Scenes

The semantic segmentation of surgical scenes is a prerequisite for task ...
research
12/03/2022

PartSLIP: Low-Shot Part Segmentation for 3D Point Clouds via Pretrained Image-Language Models

Generalizable 3D part segmentation is important but challenging in visio...
research
07/18/2023

MarS3D: A Plug-and-Play Motion-Aware Model for Semantic Segmentation on Multi-Scan 3D Point Clouds

3D semantic segmentation on multi-scan large-scale point clouds plays an...
research
06/30/2022

LaserMix for Semi-Supervised LiDAR Semantic Segmentation

Densely annotating LiDAR point clouds is costly, which restrains the sca...
research
09/08/2012

Difference of Normals as a Multi-Scale Operator in Unorganized Point Clouds

A novel multi-scale operator for unorganized 3D point clouds is introduc...
research
03/03/2021

LiDAR-based Recurrent 3D Semantic Segmentation with Temporal Memory Alignment

Understanding and interpreting a 3d environment is a key challenge for a...

Please sign up or login with your details

Forgot password? Click here to reset