Boundary Uncertainty in a Single-Stage Temporal Action Localization Network

TL;DR

This paper introduces a novel single-stage neural network for temporal action localization that models boundary uncertainties as Gaussian distributions, improving detection accuracy over previous methods.

Contribution

It is the first to model boundary uncertainties as Gaussian distributions in this context, enhancing localization performance with a simple one-stage network.

Findings

Improved mAP@tIoU=0.5 by over 1.5% using uncertainty modeling.

Boundary uncertainty modeling enhances detection accuracy.

Simple one-stage network performs comparably to complex methods.

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 $\ell_1$ loss under the same Gaussian. We show that with both uncertainty modeling approaches improve the detection performance by more than $1.5\%$ in mAP@tIoU=0.5 and that the proposed simple one-stage network performs closely to more complex one and two stage networks.