Weakly-Supervised Alignment of Video With Text

TL;DR

This paper introduces a weakly-supervised method for aligning videos with their textual descriptions by formulating the task as a temporal assignment problem and solving it efficiently, improving over previous approaches.

Contribution

It presents a novel formulation of video-text alignment as an integer quadratic program with a convex relaxation and efficient solution techniques, advancing the state of the art.

Findings

Significant improvements over previous methods on alignment tasks

Effective handling of both symbolic and continuous text representations

Robust performance on challenging video-text datasets

Abstract

Suppose that we are given a set of videos, along with natural language descriptions in the form of multiple sentences (e.g., manual annotations, movie scripts, sport summaries etc.), and that these sentences appear in the same temporal order as their visual counterparts. We propose in this paper a method for aligning the two modalities, i.e., automatically providing a time stamp for every sentence. Given vectorial features for both video and text, we propose to cast this task as a temporal assignment problem, with an implicit linear mapping between the two feature modalities. We formulate this problem as an integer quadratic program, and solve its continuous convex relaxation using an efficient conditional gradient algorithm. Several rounding procedures are proposed to construct the final integer solution. After demonstrating significant improvements over the state of the art on the related task of aligning video with symbolic labels [7], we evaluate our method on a challenging dataset of videos with associated textual descriptions [36], using both bag-of-words and continuous representations for text.