Spatio-Temporal Graph for Video Captioning with Knowledge Distillation

TL;DR

This paper introduces a spatio-temporal graph model for video captioning that captures object interactions over space and time, providing interpretable and visually grounded captions, enhanced by an object-aware knowledge distillation mechanism.

Contribution

It presents a novel spatio-temporal graph approach for video captioning with explicit visual grounding and a knowledge distillation method to stabilize performance across varying object counts.

Findings

Achieves competitive performance on benchmark datasets.

Provides interpretable and visually grounded captions.

Demonstrates effectiveness of object-aware knowledge distillation.

Abstract

Video captioning is a challenging task that requires a deep understanding of visual scenes. State-of-the-art methods generate captions using either scene-level or object-level information but without explicitly modeling object interactions. Thus, they often fail to make visually grounded predictions, and are sensitive to spurious correlations. In this paper, we propose a novel spatio-temporal graph model for video captioning that exploits object interactions in space and time. Our model builds interpretable links and is able to provide explicit visual grounding. To avoid unstable performance caused by the variable number of objects, we further propose an object-aware knowledge distillation mechanism, in which local object information is used to regularize global scene features. We demonstrate the efficacy of our approach through extensive experiments on two benchmarks, showing our approach yields competitive performance with interpretable predictions.