Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SimBase: A Simple Baseline for Temporal Video Grounding

Peijun Bao, Alex C. Kot

TL;DR

This paper designs SimBase, a network that leverages lightweight, one-dimensional temporal convolutional layers instead of complex temporal structures, and achieves state-of-the-art results on two large-scale datasets.

Abstract

This paper presents SimBase, a simple yet effective baseline for temporal video grounding. While recent advances in temporal grounding have led to impressive performance, they have also driven network architectures toward greater complexity, with a range of methods to (1) capture temporal relationships and (2) achieve effective multimodal fusion. In contrast, this paper explores the question: How effective can a simplified approach be? To investigate, we design SimBase, a network that leverages lightweight, one-dimensional temporal convolutional layers instead of complex temporal structures. For cross-modal interaction, SimBase only employs an element-wise product instead of intricate multimodal fusion. Remarkably, SimBase achieves state-of-the-art results on two large-scale datasets. As a simple yet powerful baseline, we hope SimBase will spark new ideas and streamline future evaluations in temporal video grounding.

SimBase: A Simple Baseline for Temporal Video Grounding

TL;DR

This paper designs SimBase, a network that leverages lightweight, one-dimensional temporal convolutional layers instead of complex temporal structures, and achieves state-of-the-art results on two large-scale datasets.

Abstract

This paper presents SimBase, a simple yet effective baseline for temporal video grounding. While recent advances in temporal grounding have led to impressive performance, they have also driven network architectures toward greater complexity, with a range of methods to (1) capture temporal relationships and (2) achieve effective multimodal fusion. In contrast, this paper explores the question: How effective can a simplified approach be? To investigate, we design SimBase, a network that leverages lightweight, one-dimensional temporal convolutional layers instead of complex temporal structures. For cross-modal interaction, SimBase only employs an element-wise product instead of intricate multimodal fusion. Remarkably, SimBase achieves state-of-the-art results on two large-scale datasets. As a simple yet powerful baseline, we hope SimBase will spark new ideas and streamline future evaluations in temporal video grounding.

Paper Structure

This paper contains 13 sections, 6 equations, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. SimBase
  3. Overview
  4. Temporal Interaction and Multimodal Fusion
  5. Proposal Construction
  6. Proposal Regression
  7. Training loss
  8. Experiments
  9. Datasets
  10. Evaluation Metrics
  11. Implementation Details
  12. Performance Comparisons
  13. Conclusion

Figures (2)

  • Figure 1: Examples of temporal video grounding. The goal of temporal video grounding is to localize the temporal moment in an untrimmed video based on a given sentence query.
  • Figure 2: Overview of Simple Baseline (SimBase) for temporal video grounding. SimBase leverages lightweight, one-dimensional temporal convolutional layers instead of complex temporal structures. For cross-modal interaction, SimBase exploits Hadamard product rather than complex interaction of language and video.