BLIP-2

BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models

摘要

The cost of vision-and-language pre-training has become increasingly prohibitive due to end-toend training of large-scale models. This paper proposes BLIP-2, a generic and efficient pretraining strategy that bootstraps vision-language pre-training from off-the-shelf frozen pre-trained image encoders and frozen large language models. BLIP-2 bridges the modality gap with a lightweight Querying Transformer, which is pretrained in two stages. The first stage bootstraps vision-language representation learning from a frozen image encoder. The second stage bootstraps vision-to-language generative learning from a frozen language model. BLIP-2 achieves state-of-the-art performance on various visionlanguage tasks, despite having significantly fewer trainable parameters than existing methods. For example, our model outperforms Flamingo80B by 8.7% on zero-shot VQAv2 with 54x fewer trainable parameters. We also demonstrate the model’s emerging capabilities of zero-shot image-to-text generation that can follow natural language instructions.

使用方式

Use the modelTest Command

from mmpretrain import inference_model

result = inference_model('blip2-opt2.7b_3rdparty-zeroshot_caption', 'demo/cat-dog.png')
print(result)
# {'pred_caption': 'a dog and a cat sitting on a blanket'}

Models and results

Image Caption on COCO

模型	Params (M)	BLEU-4	CIDER	配置文件	下载
blip2-opt2.7b_3rdparty-zeroshot_caption*	3770.47	32.90	111.10	config	model

Visual Question Answering on VQAv2

模型	Params (M)	Accuracy	配置文件	下载
blip2-opt2.7b_3rdparty-zeroshot_vqa*	3770.47	53.50	config	model

Image-To-Text Retrieval on COCO

模型	Params (M)	Recall@1	配置文件	下载
blip2_3rdparty_retrieval*	1173.19	85.40	config	model

Models with * are converted from the official repo. The config files of these models are only for inference. We haven’t reproduce the training results.

引用

@article{beitv2,
    title={Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models},
    author={Li, Junnan and Li, Dongxu and Savarese, Silvio and Hoi, Steven},
    year={2023},
    eprint={2301.12597},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}