Non-Autoregressive Coarse-to-Fine Video Captioning

TL;DR

This paper introduces a non-autoregressive, coarse-to-fine model for video captioning that significantly speeds up inference and improves description diversity and accuracy by focusing on visual words and iterative refinement.

Contribution

It proposes a novel non-autoregressive, coarse-to-fine captioning framework with a bi-directional self-attention model and a visual word generation mechanism, achieving state-of-the-art results.

Findings

Achieves faster inference speed compared to autoregressive models.

Generates more diverse and accurate video descriptions.

Outperforms existing methods on MSVD and MSR-VTT benchmarks.

Abstract

It is encouraged to see that progress has been made to bridge videos and natural language. However, mainstream video captioning methods suffer from slow inference speed due to the sequential manner of autoregressive decoding, and prefer generating generic descriptions due to the insufficient training of visual words (e.g., nouns and verbs) and inadequate decoding paradigm. In this paper, we propose a non-autoregressive decoding based model with a coarse-to-fine captioning procedure to alleviate these defects. In implementations, we employ a bi-directional self-attention based network as our language model for achieving inference speedup, based on which we decompose the captioning procedure into two stages, where the model has different focuses. Specifically, given that visual words determine the semantic correctness of captions, we design a mechanism of generating visual words to not only promote the training of scene-related words but also capture relevant details from videos to construct a coarse-grained sentence "template". Thereafter, we devise dedicated decoding algorithms that fill in the "template" with suitable words and modify inappropriate phrasing via iterative refinement to obtain a fine-grained description. Extensive experiments on two mainstream video captioning benchmarks, i.e., MSVD and MSR-VTT, demonstrate that our approach achieves state-of-the-art performance, generates diverse descriptions, and obtains high inference efficiency. Our code is available at https://github.com/yangbang18/Non-Autoregressive-Video-Captioning.