Transform and Tell: Entity-Aware News Image Captioning

英文摘要：

We propose an end-to-end model which generates captions for images embedded in news articles. News images present two key challenges: they rely on real-world knowledge, especially about named entities; and they typically have linguistically rich captions that include uncommon words. We address the first challenge by associating words in the caption with faces and objects in the image, via a multi-modal, multi-head attention mechanism. We tackle the second challenge with a state-of-the-art transformer language model that uses byte-pair-encoding to generate captions as a sequence of word parts. On the GoodNews dataset, our model outperforms the previous state of the art by a factor of four in CIDEr score (13 to 54). This performance gain comes from a unique combination of language models, word representation, image embeddings, face embeddings, object embeddings, and improvements in neural network design. We also introduce the NYTimes800k dataset which is 70% larger than GoodNews, has higher article quality, and includes the locations of images within articles as an additional contextual cue.