Metraipolis › archive › dc_bc74681f5d43

PT$^3$: Prototype-Guided Self-Supervised Pretraining for Generalizable mmWave Human Point Cloud Representation — arXiv:2610.38551

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.38551
License (as archived)
CC0-1.0
Archived
2026-10-31T06:08:58.888Z
SHA-256 of the archived text
f497d57ec4a77b3644b490b86a148f64c0c2f7e88f9237571ae16febc4cbd730
Stable URL of this page
https://metraipolis.com/documento/dc_bc74681f5d43

How to cite

PT$^3$: Prototype-Guided Self-Supervised Pretraining for Generalizable mmWave Human Point Cloud Representation — arXiv:2610.38551 — arXiv (en), https://arxiv.org/abs/2610.38551 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bc74681f5d43 (Metraipolis archive dc_bc74681f5d43, sha256 f497d57ec4a77b3644b490b86a148f64c0c2f7e88f9237571ae16febc4cbd730)
BibTeX
@misc{dcbc74681f5d43,
  title = {PT$^3$: Prototype-Guided Self-Supervised Pretraining for Generalizable mmWave Human Point Cloud Representation — arXiv:2610.38551},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.38551},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bc74681f5d43; sha256 f497d57ec4a77b3644b490b86a148f64c0c2f7e88f9237571ae16febc4cbd730},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.38551. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multi-source millimeter-wave (mmWave) observations offer an opportunity to learn human representations reusable across sensing tasks. However, sparse detections and multipath effects make these observations unstable, challenging existing point cloud pretraining methods. We propose PT$^3$, a prototype-guided self-supervised pretraining framework for mmWave human point clouds. PT$^3$ represents local states in spatial regions as sparse compositions of shared prototypes and learns their temporal evolution through autoregressive prediction. Observation consistency further stabilizes representations under radar perturbations. After pretraining on unlabeled sequences from multiple sources, the same frozen encoder supports action classification, dense point generation, joint pose estimation, and future motion prediction on datasets not used for pretraining. Across these tasks, PT$^3$ outperforms the compared pretraining baselines and achieves competitive or superior performance to task-specific models.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.38551 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-31T06:07:08.482Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).