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

Junnan Li, Dongxu Li, Caiming Xiong, Steven Hoi
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:12888-12900, 2022.

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 and models are available at https://github.com/salesforce/BLIP.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-li22n, title = {{BLIP}: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation}, author = {Li, Junnan and Li, Dongxu and Xiong, Caiming and Hoi, Steven}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {12888--12900}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/li22n/li22n.pdf}, url = {https://proceedings.mlr.press/v162/li22n.html}, 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 and models are available at https://github.com/salesforce/BLIP.} }
Endnote
%0 Conference Paper %T BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation %A Junnan Li %A Dongxu Li %A Caiming Xiong %A Steven Hoi %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-li22n %I PMLR %P 12888--12900 %U https://proceedings.mlr.press/v162/li22n.html %V 162 %X 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 and models are available at https://github.com/salesforce/BLIP.
APA
Li, J., Li, D., Xiong, C. & Hoi, S.. (2022). BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:12888-12900 Available from https://proceedings.mlr.press/v162/li22n.html.

Related Material