Topology-Guided Knowledge Distillation for Efficient Point Cloud Processing

TL;DR

This paper presents a topology-aware knowledge distillation framework that significantly reduces model size and inference time for point cloud processing while maintaining high performance, enabling efficient deployment in resource-limited environments.

Contribution

It introduces a novel topology-guided distillation method that captures geometric structures and uses gradient-based feature alignment for improved point cloud model compression.

Findings

Achieves 16x smaller models with 1.9x faster inference.

Surpasses prior distillation methods on NuScenes segmentation.

Maintains competitive performance across multiple datasets.

Abstract

Point cloud processing has gained significant attention due to its critical role in applications such as autonomous driving and 3D object recognition. However, deploying high-performance models like Point Transformer V3 in resource-constrained environments remains challenging due to their high computational and memory demands. This work introduces a novel distillation framework that leverages topology-aware representations and gradient-guided knowledge distillation to effectively transfer knowledge from a high-capacity teacher to a lightweight student model. Our approach captures the underlying geometric structures of point clouds while selectively guiding the student model's learning process through gradient-based feature alignment. Experimental results in the Nuscenes, SemanticKITTI, and Waymo datasets demonstrate that the proposed method achieves competitive performance, with an approximately 16x reduction in model size and a nearly 1.9x decrease in inference time compared to its teacher model. Notably, on NuScenes, our method achieves state-of-the-art performance among knowledge distillation techniques trained solely on LiDAR data, surpassing prior knowledge distillation baselines in segmentation performance. Our implementation is available publicly at: https://github.com/HySonLab/PointDistill