English

Boundary Uncertainty in a Single-Stage Temporal Action Localization Network

Computer Vision and Pattern Recognition 2020-08-26 v1

Abstract

In this paper, we address the problem of temporal action localization with a single stage neural network. In the proposed architecture we model the boundary predictions as uni-variate Gaussian distributions in order to model their uncertainties, which is the first in this area to the best of our knowledge. We use two uncertainty-aware boundary regression losses: first, the Kullback-Leibler divergence between the ground truth location of the boundary and the Gaussian modeling the prediction of the boundary and second, the expectation of the 1\ell_1 loss under the same Gaussian. We show that with both uncertainty modeling approaches improve the detection performance by more than 1.5%1.5\% in mAP@tIoU=0.5 and that the proposed simple one-stage network performs closely to more complex one and two stage networks.

Cite

@article{arxiv.2008.11170,
  title  = {Boundary Uncertainty in a Single-Stage Temporal Action Localization Network},
  author = {Ting-Ting Xie and Christos Tzelepis and Ioannis Patras},
  journal= {arXiv preprint arXiv:2008.11170},
  year   = {2020}
}

Comments

Tech report

R2 v1 2026-06-23T18:05:53.816Z