FastPoint: Accelerating 3D Point Cloud Model Inference via Sample Point Distance Prediction

Donghyun Lee, Dawoon Jeong, Jae W. Lee, Hongil Yoon · ICCV 2025 · 2025

DOI arXiv

Status

To read

Topics

sampling, budgeted-inference

Feeds

P1 · ICCV 2027

What they did

Accelerates farthest point sampling and neighbour search by predicting how the distance between successive sample points decreases, so later samples are found without computing all pairwise distances; sampling quality and accuracy are kept.

My note

Accelerates farthest-point sampling and neighborhood operations. Helps distinguish better sampling decisions from faster sampling implementations.

BibTeX

@inproceedings{lee2025-fastpoint,
  title = {{FastPoint: Accelerating 3D Point Cloud Model Inference via Sample Point Distance Prediction}},
  author = {Donghyun Lee and Dawoon Jeong and Jae W. Lee and Hongil Yoon},
  booktitle = {ICCV 2025},
  year = {2025},
  doi = {10.1109/ICCV51701.2025.02329},
  eprint = {2507.23480},
  archivePrefix = {arXiv}
}