LOGen: Toward Lidar Object Generation by Point Diffusion

2024-12-10

arXiv.org

AI Summary

To address the data scarcity of rare semantic classes (e.g., pedestrians) in LiDAR-based autonomous driving, this work pioneers the application of diffusion models to object-level point cloud generation. We propose a novel point cloud diffusion framework enabling joint reflectance-geometry modeling and multi-condition controllable generation—supporting class label, 3D bounding box dimensions, and orientation as explicit conditioning inputs. The method introduces a lightweight conditional embedding mechanism and a reflectance-aware denoising network. Furthermore, we establish the first comprehensive 3D evaluation metric suite specifically designed for LiDAR object generation, quantifying geometric fidelity, semantic consistency, and diversity. Extensive experiments on the nuScenes dataset demonstrate that our approach significantly outperforms existing baselines, achieving superior quality, high geometric and semantic fidelity, and strong controllability in generating vehicle and pedestrian point clouds.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

A common strategy to improve lidar segmentation results on rare semantic classes consists of pasting objects from one lidar scene into another. While this augments the quantity of instances seen at training time and varies their context, the instances fundamentally remain the same. In this work, we explore how to enhance instance diversity using a lidar object generator. We introduce a novel diffusion-based method to produce lidar point clouds of dataset objects, including reflectance, and with an extensive control of the generation via conditioning information. Our experiments on nuScenes show the quality of our object generations measured with new 3D metrics developed to suit lidar objects.

Problem

• Generating realistic LiDAR scans for autonomous driving applications.
• Focusing on 3D object generation as viewed by LiDAR sensors.
• Developing a diffusion-based model for controlled LiDAR point cloud generation.

Innovation

• Diffusion-based model for LiDAR point clouds
• Conditioning information controls generation
• New 3D metrics for LiDAR object quality