2D Gaussians Spatial Transport for Point-supervised Density Regression

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shang, Miao, Hong, Xiaopeng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917099717263360
author Shang, Miao
Hong, Xiaopeng
author_facet Shang, Miao
Hong, Xiaopeng
contents This paper introduces Gaussian Spatial Transport (GST), a novel framework that leverages Gaussian splatting to facilitate transport from the probability measure in the image coordinate space to the annotation map. We propose a Gaussian splatting-based method to estimate pixel-annotation correspondence, which is then used to compute a transport plan derived from Bayesian probability. To integrate the resulting transport plan into standard network optimization in typical computer vision tasks, we derive a loss function that measures discrepancy after transport. Extensive experiments on representative computer vision tasks, including crowd counting and landmark detection, validate the effectiveness of our approach. Compared to conventional optimal transport schemes, GST eliminates iterative transport plan computation during training, significantly improving efficiency. Code is available at https://github.com/infinite0522/GST.
format Preprint
id arxiv_https___arxiv_org_abs_2511_14477
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle 2D Gaussians Spatial Transport for Point-supervised Density Regression
Shang, Miao
Hong, Xiaopeng
Computer Vision and Pattern Recognition
This paper introduces Gaussian Spatial Transport (GST), a novel framework that leverages Gaussian splatting to facilitate transport from the probability measure in the image coordinate space to the annotation map. We propose a Gaussian splatting-based method to estimate pixel-annotation correspondence, which is then used to compute a transport plan derived from Bayesian probability. To integrate the resulting transport plan into standard network optimization in typical computer vision tasks, we derive a loss function that measures discrepancy after transport. Extensive experiments on representative computer vision tasks, including crowd counting and landmark detection, validate the effectiveness of our approach. Compared to conventional optimal transport schemes, GST eliminates iterative transport plan computation during training, significantly improving efficiency. Code is available at https://github.com/infinite0522/GST.
title 2D Gaussians Spatial Transport for Point-supervised Density Regression
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.14477