Reconstruction Network for Video Captioning

TL;DR

This paper introduces RecNet, a novel encoder-decoder-reconstructor architecture for video captioning that uses bidirectional flows to improve caption accuracy by jointly training with generation and reconstruction losses.

Contribution

The paper proposes a reconstruction network with a bidirectional flow approach, enhancing video captioning by jointly optimizing caption generation and video feature reconstruction.

Findings

Significant improvement in caption accuracy on benchmark datasets.

Reconstruction loss boosts encoder-decoder model performance.

Bidirectional flow approach effectively leverages video and sentence information.

Abstract

In this paper, the problem of describing visual contents of a video sequence with natural language is addressed. Unlike previous video captioning work mainly exploiting the cues of video contents to make a language description, we propose a reconstruction network (RecNet) with a novel encoder-decoder-reconstructor architecture, which leverages both the forward (video to sentence) and backward (sentence to video) flows for video captioning. Specifically, the encoder-decoder makes use of the forward flow to produce the sentence description based on the encoded video semantic features. Two types of reconstructors are customized to employ the backward flow and reproduce the video features based on the hidden state sequence generated by the decoder. The generation loss yielded by the encoder-decoder and the reconstruction loss introduced by the reconstructor are jointly drawn into training the proposed RecNet in an end-to-end fashion. Experimental results on benchmark datasets demonstrate that the proposed reconstructor can boost the encoder-decoder models and leads to significant gains in video caption accuracy.