MICAS: Multi-grained In-Context Adaptive Sampling for 3D Point Cloud Processing

Feifei Shao, Ping Liu, Zhao Wang, Yawei Luo, Hongwei Wang, Jun Xiao · CVPR 2025 · 2025

DOI arXiv

Status

To read

Topics

sampling

Feeds

P1 · ICCV 2027

What they did

Proposes MICAS, an in-context learning framework for point cloud processing that adapts sampling at two levels: task-adaptive point sampling and query-specific prompt sampling.

My note

Combines task-adaptive point sampling and query-specific prompt selection. Relevant precedent for task-conditioned selection.

BibTeX

@inproceedings{shao2025-micas,
  title = {{MICAS: Multi-grained In-Context Adaptive Sampling for 3D Point Cloud Processing}},
  author = {Feifei Shao and Ping Liu and Zhao Wang and Yawei Luo and Hongwei Wang and Jun Xiao},
  booktitle = {CVPR 2025},
  year = {2025},
  doi = {10.1109/CVPR52734.2025.00620},
  eprint = {2411.16773},
  archivePrefix = {arXiv}
}