Contrastive Video Question Answering via Video Graph Transformer

TL;DR

This paper introduces CoVGT, a novel contrastive video question answering model that uses a video graph transformer to improve fine-grained reasoning and outperforms previous methods with less data.

Contribution

It presents a dynamic graph transformer for detailed video encoding and contrastive learning for video-text matching, advancing VideoQA performance and data efficiency.

Findings

CoVGT surpasses previous models on video reasoning tasks.

It achieves high performance with significantly less data.

The model benefits from cross-modal pretraining.

Abstract

We propose to perform video question answering (VideoQA) in a Contrastive manner via a Video Graph Transformer model (CoVGT). CoVGT's uniqueness and superiority are three-fold: 1) It proposes a dynamic graph transformer module which encodes video by explicitly capturing the visual objects, their relations and dynamics, for complex spatio-temporal reasoning. 2) It designs separate video and text transformers for contrastive learning between the video and text to perform QA, instead of multi-modal transformer for answer classification. Fine-grained video-text communication is done by additional cross-modal interaction modules. 3) It is optimized by the joint fully- and self-supervised contrastive objectives between the correct and incorrect answers, as well as the relevant and irrelevant questions respectively. With superior video encoding and QA solution, we show that CoVGT can achieve much better performances than previous arts on video reasoning tasks. Its performances even surpass those models that are pretrained with millions of external data. We further show that CoVGT can also benefit from cross-modal pretraining, yet with orders of magnitude smaller data. The results demonstrate the effectiveness and superiority of CoVGT, and additionally reveal its potential for more data-efficient pretraining. We hope our success can advance VideoQA beyond coarse recognition/description towards fine-grained relation reasoning of video contents. Our code is available at https://github.com/doc-doc/CoVGT.