Attention on Attention for Image Captioning

TL;DR

This paper introduces the Attention on Attention (AoA) module, enhancing image captioning models by better assessing attention relevance, leading to state-of-the-art performance on MS COCO dataset.

Contribution

The paper proposes the AoA module that extends traditional attention mechanisms to improve relevance estimation, applied to both encoder and decoder in image captioning.

Findings

Achieved a CIDEr-D score of 129.8 on MS COCO test split.

Outperformed all previous methods in image captioning.

Demonstrated the effectiveness of AoA in both encoder and decoder.

Abstract

Attention mechanisms are widely used in current encoder/decoder frameworks of image captioning, where a weighted average on encoded vectors is generated at each time step to guide the caption decoding process. However, the decoder has little idea of whether or how well the attended vector and the given attention query are related, which could make the decoder give misled results. In this paper, we propose an Attention on Attention (AoA) module, which extends the conventional attention mechanisms to determine the relevance between attention results and queries. AoA first generates an information vector and an attention gate using the attention result and the current context, then adds another attention by applying element-wise multiplication to them and finally obtains the attended information, the expected useful knowledge. We apply AoA to both the encoder and the decoder of our image captioning model, which we name as AoA Network (AoANet). Experiments show that AoANet outperforms all previously published methods and achieves a new state-of-the-art performance of 129.8 CIDEr-D score on MS COCO Karpathy offline test split and 129.6 CIDEr-D (C40) score on the official online testing server. Code is available at https://github.com/husthuaan/AoANet.