Dynamic Convolution for 3D Point Cloud Instance Segmentation

TL;DR

This paper introduces a dynamic convolution approach for 3D point cloud instance segmentation that adapts to varying object scales and improves representation, resulting in robust and efficient performance across multiple datasets.

Contribution

It presents a proposal-free, adaptive convolution method with a lightweight transformer for enhanced 3D point cloud instance segmentation.

Findings

Achieves strong performance on ScanNetV2, S3DIS, and PartNet datasets.

Improves robustness and efficiency over existing methods.

Demonstrates effectiveness across voxel- and point-based architectures.

Abstract

We propose an approach to instance segmentation from 3D point clouds based on dynamic convolution. This enables it to adapt, at inference, to varying feature and object scales. Doing so avoids some pitfalls of bottom up approaches, including a dependence on hyper-parameter tuning and heuristic post-processing pipelines to compensate for the inevitable variability in object sizes, even within a single scene. The representation capability of the network is greatly improved by gathering homogeneous points that have identical semantic categories and close votes for the geometric centroids. Instances are then decoded via several simple convolution layers, where the parameters are generated conditioned on the input. The proposed approach is proposal-free, and instead exploits a convolution process that adapts to the spatial and semantic characteristics of each instance. A light-weight transformer, built on the bottleneck layer, allows the model to capture long-range dependencies, with limited computational overhead. The result is a simple, efficient, and robust approach that yields strong performance on various datasets: ScanNetV2, S3DIS, and PartNet. The consistent improvements on both voxel- and point-based architectures imply the effectiveness of the proposed method. Code is available at: https://git.io/DyCo3D