# DOS: Distilling Observable Softmaps of Zipfian Prototypes for Self-Supervised Point Representation

arXiv: https://arxiv.org/abs/2512.11465 · HTML: https://mlsanity.com/paper/2512.11465/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 28 references that are in the map.
- First posted on arXiv 12 Dec 2025.

## Builds on (most cited first, 16 of 16)

- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization](https://mlsanity.com/paper/1610.02391/) (2016)
- [ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes](https://mlsanity.com/paper/1702.04405/) (2017)
- [Unsupervised Learning of Visual Features by Contrasting Cluster Assignments](https://mlsanity.com/paper/2006.09882/) (2020)
- [Scalability in Perception for Autonomous Driving: Waymo Open Dataset](https://mlsanity.com/paper/1912.04838/) (2019)
- [Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modeling](https://mlsanity.com/paper/2111.14819/) (2021)
- [Point Transformer V3: Simpler, Faster, Stronger](https://mlsanity.com/paper/2312.10035/) (2023)
- [Few-shot Object Detection via Feature Reweighting](https://mlsanity.com/paper/1812.01866/) (2018)
- [UniPAD: A Universal Pre-training Paradigm for Autonomous Driving](https://mlsanity.com/paper/2310.08370/) (2023)
- [Learning from 2D: Contrastive Pixel-to-Point Knowledge Transfer for 3D Pretraining](https://mlsanity.com/paper/2104.04687/) (2021)
- [DINO in the Room: Leveraging 2D Foundation Models for 3D Segmentation](https://mlsanity.com/paper/2503.18944/) (2025)
- [A Point-Based Approach to Efficient LiDAR Multi-Task Perception](https://mlsanity.com/paper/2404.12798/) (2024)
- [Label-Efficient LiDAR Semantic Segmentation with 2D-3D Vision Transformer Adapters](https://mlsanity.com/paper/2503.03299/) (2025)
- [Self-Supervised Multi-Object Tracking For Autonomous Driving From Consistency Across Timescales](https://mlsanity.com/paper/2304.13147/) (2023)
- [Open-Set LiDAR Panoptic Segmentation Guided by Uncertainty-Aware Learning](https://mlsanity.com/paper/2506.13265/) (2025)
- [ParisLuco3D: A high-quality target dataset for domain generalization of LiDAR perception](https://mlsanity.com/paper/2310.16542/) (2023)

## Built on by (most cited first, 2 of 2)

- [Towards Foundation Models for 3D Scene Understanding: Instance-Aware Self-Supervised Learning for Point Clouds](https://mlsanity.com/paper/2603.25165/) (2026)
- [Collaborative Learning for Semi-Supervised LiDAR Semantic Segmentation](https://mlsanity.com/paper/2605.17135/) (2026)

## Abstract (arXiv)

Recent advances in self-supervised learning (SSL) have shown tremendous potential for learning 3D point cloud representations without human annotations. However, SSL for 3D point clouds still faces critical challenges due to irregular geometry, shortcut-prone reconstruction, and unbalanced semantics distribution. In this work, we propose DOS (Distilling Observable Softmaps), a novel SSL framework that self-distills semantic relevance softmaps only at observable (unmasked) points. This strategy prevents information leakage from masked regions and provides richer supervision than discrete token-to-prototype assignments. To address the challenge of unbalanced semantics in an unsupervised setting, we introduce Zipfian prototypes and incorporate them using a modified Sinkhorn-Knopp algorithm, Zipf-Sinkhorn, which enforces a power-law prior over prototype usage and modulates the sharpness of the target softmap during training. DOS outperforms current state-of-the-art methods on semantic segmentation and 3D object detection across multiple benchmarks, including nuScenes, Waymo, SemanticKITTI, ScanNet, and ScanNet200, without relying on extra data or annotations. Our results demonstrate that observable-point softmaps distillation offers a scalable and effective paradigm for learning robust 3D representations.