BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation

TL;DR

BLIP introduces a flexible vision-language pre-training framework that leverages noisy web data through bootstrapping, achieving state-of-the-art results across understanding and generation tasks, and demonstrating strong zero-shot transfer to video-language tasks.

Contribution

BLIP presents a novel bootstrapping approach for VLP that effectively utilizes noisy web data and unifies vision-language understanding and generation tasks.

Findings

State-of-the-art results on image-text retrieval, captioning, and VQA.

Effective use of noisy web data via caption generation and filtering.

Strong zero-shot transfer performance to video-language tasks.

Abstract

Vision-Language Pre-training (VLP) has advanced the performance for many vision-language tasks. However, most existing pre-trained models only excel in either understanding-based tasks or generation-based tasks. Furthermore, performance improvement has been largely achieved by scaling up the dataset with noisy image-text pairs collected from the web, which is a suboptimal source of supervision. In this paper, we propose BLIP, a new VLP framework which transfers flexibly to both vision-language understanding and generation tasks. BLIP effectively utilizes the noisy web data by bootstrapping the captions, where a captioner generates synthetic captions and a filter removes the noisy ones. We achieve state-of-the-art results on a wide range of vision-language tasks, such as image-text retrieval (+2.7% in average recall@1), image captioning (+2.8% in CIDEr), and VQA (+1.6% in VQA score). BLIP also demonstrates strong generalization ability when directly transferred to video-language tasks in a zero-shot manner. Code, models, and datasets are released at https://github.com/salesforce/BLIP.