Unifying Visual-Semantic Embeddings with Multimodal Neural Language Models

TL;DR

This paper presents a unified multimodal neural model that learns joint image-text embeddings and generates descriptions, achieving state-of-the-art results on benchmark datasets without relying on object detections.

Contribution

It introduces a novel encoder-decoder framework that combines joint embedding learning with a neural language model, unifying image-text retrieval and caption generation.

Findings

Achieves state-of-the-art performance on Flickr8K and Flickr30K datasets.

Demonstrates that the embedding space captures multimodal regularities like vector arithmetic.

Sets new benchmarks with deeper convolutional networks.

Abstract

Inspired by recent advances in multimodal learning and machine translation, we introduce an encoder-decoder pipeline that learns (a): a multimodal joint embedding space with images and text and (b): a novel language model for decoding distributed representations from our space. Our pipeline effectively unifies joint image-text embedding models with multimodal neural language models. We introduce the structure-content neural language model that disentangles the structure of a sentence to its content, conditioned on representations produced by the encoder. The encoder allows one to rank images and sentences while the decoder can generate novel descriptions from scratch. Using LSTM to encode sentences, we match the state-of-the-art performance on Flickr8K and Flickr30K without using object detections. We also set new best results when using the 19-layer Oxford convolutional network. Furthermore we show that with linear encoders, the learned embedding space captures multimodal regularities in terms of vector space arithmetic e.g. *image of a blue car* - "blue" + "red" is near images of red cars. Sample captions generated for 800 images are made available for comparison.