LightningDOT: Pre-training Visual-Semantic Embeddings for Real-Time Image-Text Retrieval

TL;DR

LightningDOT is a novel approach that drastically accelerates image-text retrieval by removing cross-modal attention during inference, achieving state-of-the-art accuracy with significantly reduced computation and latency.

Contribution

It introduces a pre-training method that enables instant dot-product matching for image-text retrieval, eliminating the need for slow cross-modal attention during inference.

Findings

Achieves thousands of times faster inference speed.

Outperforms existing models on Flickr30k, COCO, and Multi30K benchmarks.

Maintains high accuracy comparable to state-of-the-art methods.

Abstract

Multimodal pre-training has propelled great advancement in vision-and-language research. These large-scale pre-trained models, although successful, fatefully suffer from slow inference speed due to enormous computation cost mainly from cross-modal attention in Transformer architecture. When applied to real-life applications, such latency and computation demand severely deter the practical use of pre-trained models. In this paper, we study Image-text retrieval (ITR), the most mature scenario of V+L application, which has been widely studied even prior to the emergence of recent pre-trained models. We propose a simple yet highly effective approach, LightningDOT that accelerates the inference time of ITR by thousands of times, without sacrificing accuracy. LightningDOT removes the time-consuming cross-modal attention by pre-training on three novel learning objectives, extracting feature indexes offline, and employing instant dot-product matching with further re-ranking, which significantly speeds up retrieval process. In fact, LightningDOT achieves new state of the art across multiple ITR benchmarks such as Flickr30k, COCO and Multi30K, outperforming existing pre-trained models that consume 1000x magnitude of computational hours. Code and pre-training checkpoints are available at https://github.com/intersun/LightningDOT.