A Neural Compositional Paradigm for Image Captioning

TL;DR

This paper introduces a two-stage, compositional approach to image captioning that explicitly separates semantic extraction from caption generation, resulting in more accurate, diverse, and generalizable captions.

Contribution

It proposes a novel paradigm that factorizes captioning into semantic extraction and recursive compositional caption construction, improving diversity and generalization over traditional sequential models.

Findings

Better preservation of semantic content

Requires less training data

Produces more diverse captions

Abstract

Mainstream captioning models often follow a sequential structure to generate captions, leading to issues such as introduction of irrelevant semantics, lack of diversity in the generated captions, and inadequate generalization performance. In this paper, we present an alternative paradigm for image captioning, which factorizes the captioning procedure into two stages: (1) extracting an explicit semantic representation from the given image; and (2) constructing the caption based on a recursive compositional procedure in a bottom-up manner. Compared to conventional ones, our paradigm better preserves the semantic content through an explicit factorization of semantics and syntax. By using the compositional generation procedure, caption construction follows a recursive structure, which naturally fits the properties of human language. Moreover, the proposed compositional procedure requires less data to train, generalizes better, and yields more diverse captions.