Attention Beam: An Image Captioning Approach

TL;DR

This paper introduces an improved beam search heuristic applied to encoder-decoder models, significantly enhancing image captioning quality across multiple benchmark datasets.

Contribution

It proposes a novel beam search heuristic that improves caption quality in encoder-decoder image captioning models.

Findings

Better caption quality on Flickr8k, Flickr30k, MS COCO datasets

Outperforms existing methods in benchmark evaluations

Demonstrates effectiveness of heuristic beam search

Abstract

The aim of image captioning is to generate textual description of a given image. Though seemingly an easy task for humans, it is challenging for machines as it requires the ability to comprehend the image (computer vision) and consequently generate a human-like description for the image (natural language understanding). In recent times, encoder-decoder based architectures have achieved state-of-the-art results for image captioning. Here, we present a heuristic of beam search on top of the encoder-decoder based architecture that gives better quality captions on three benchmark datasets: Flickr8k, Flickr30k and MS COCO.