Scalable 3D Panoptic Segmentation As Superpoint Graph Clustering

TL;DR

This paper presents SuperCluster, a scalable and efficient graph clustering method for 3D panoptic segmentation that achieves state-of-the-art results on multiple large-scale datasets with significantly fewer parameters.

Contribution

It introduces a novel graph clustering approach for 3D segmentation that is resource-efficient, adaptable to superpoints, and capable of processing scenes with millions of points in real-time.

Findings

Achieves new state-of-the-art PQ scores on S3DIS and ScanNetV2 datasets.

Sets first SOTA results on KITTI-360 and DALES benchmarks.

Model is over 30 times smaller and trains 15 times faster than previous methods.

Abstract

We introduce a highly efficient method for panoptic segmentation of large 3D point clouds by redefining this task as a scalable graph clustering problem. This approach can be trained using only local auxiliary tasks, thereby eliminating the resource-intensive instance-matching step during training. Moreover, our formulation can easily be adapted to the superpoint paradigm, further increasing its efficiency. This allows our model to process scenes with millions of points and thousands of objects in a single inference. Our method, called SuperCluster, achieves a new state-of-the-art panoptic segmentation performance for two indoor scanning datasets: $50.1$ PQ ($+7.8$) for S3DIS Area~5, and $58.7$ PQ ($+25.2$) for ScanNetV2. We also set the first state-of-the-art for two large-scale mobile mapping benchmarks: KITTI-360 and DALES. With only $209$k parameters, our model is over $30$ times smaller than the best-competing method and trains up to $15$ times faster. Our code and pretrained models are available at https://github.com/drprojects/superpoint_transformer.