Gespeichert in:
| Hauptverfasser: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2503.16429 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913748305838080 |
|---|---|
| author | Wu, Xiaoyang DeTone, Daniel Frost, Duncan Shen, Tianwei Xie, Chris Yang, Nan Engel, Jakob Newcombe, Richard Zhao, Hengshuang Straub, Julian |
| author_facet | Wu, Xiaoyang DeTone, Daniel Frost, Duncan Shen, Tianwei Xie, Chris Yang, Nan Engel, Jakob Newcombe, Richard Zhao, Hengshuang Straub, Julian |
| contents | In this paper, we question whether we have a reliable self-supervised point cloud model that can be used for diverse 3D tasks via simple linear probing, even with limited data and minimal computation. We find that existing 3D self-supervised learning approaches fall short when evaluated on representation quality through linear probing. We hypothesize that this is due to what we term the "geometric shortcut", which causes representations to collapse to low-level spatial features. This challenge is unique to 3D and arises from the sparse nature of point cloud data. We address it through two key strategies: obscuring spatial information and enhancing the reliance on input features, ultimately composing a Sonata of 140k point clouds through self-distillation. Sonata is simple and intuitive, yet its learned representations are strong and reliable: zero-shot visualizations demonstrate semantic grouping, alongside strong spatial reasoning through nearest-neighbor relationships. Sonata demonstrates exceptional parameter and data efficiency, tripling linear probing accuracy (from 21.8% to 72.5%) on ScanNet and nearly doubling performance with only 1% of the data compared to previous approaches. Full fine-tuning further advances SOTA across both 3D indoor and outdoor perception tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_16429 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Sonata: Self-Supervised Learning of Reliable Point Representations Wu, Xiaoyang DeTone, Daniel Frost, Duncan Shen, Tianwei Xie, Chris Yang, Nan Engel, Jakob Newcombe, Richard Zhao, Hengshuang Straub, Julian Computer Vision and Pattern Recognition In this paper, we question whether we have a reliable self-supervised point cloud model that can be used for diverse 3D tasks via simple linear probing, even with limited data and minimal computation. We find that existing 3D self-supervised learning approaches fall short when evaluated on representation quality through linear probing. We hypothesize that this is due to what we term the "geometric shortcut", which causes representations to collapse to low-level spatial features. This challenge is unique to 3D and arises from the sparse nature of point cloud data. We address it through two key strategies: obscuring spatial information and enhancing the reliance on input features, ultimately composing a Sonata of 140k point clouds through self-distillation. Sonata is simple and intuitive, yet its learned representations are strong and reliable: zero-shot visualizations demonstrate semantic grouping, alongside strong spatial reasoning through nearest-neighbor relationships. Sonata demonstrates exceptional parameter and data efficiency, tripling linear probing accuracy (from 21.8% to 72.5%) on ScanNet and nearly doubling performance with only 1% of the data compared to previous approaches. Full fine-tuning further advances SOTA across both 3D indoor and outdoor perception tasks. |
| title | Sonata: Self-Supervised Learning of Reliable Point Representations |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2503.16429 |