Home

Awesome

<img src="./docs/logo.svg" width="150" >

This repository contains source code necessary to reproduce the results presented in the paper ZeroVL: A Strong Baseline for Aligning Vision-Language Representations with Limited Resources.

Pioneering dual-encoder pre-training works (e.g., CLIP and ALIGN) require a tremendous amount of data and computational resources (e.g., billion-level web data and hundreds of GPUs), which prevent researchers with limited resources from reproduction and further exploration. To this end, we provide a comprehensive training guidance, which allows us to conduct dual-encoder multi-modal representation alignment with limited resources. Meanwhile, we provide a reproducible strong baseline of competitive results, namely ZeroVL, with publicly accessible academic datasets and a popular experimental environment.

Performance

Image-text retreival RSUM scores on MSCOCO and Flickr30K datasets:

methodcomputationdataCOCO(zs.)COCO(ft.)F30K(zs.)F30K(ft.)
CLIP256 V100400M400.2-540.6-
ALIGN1024 TPUv31800M425.3500.4553.3576.0
baseline8 V10014.2M363.5471.9476.8553.0
ZeroVL8 V10014.2M425.0485.0536.2561.6
ZeroVL8 V100100M442.1500.5546.5573.6

zs.: zero-shot setting, ft.: fine-tuned setting.

ImageNet-1K linear probing results:

methoddatabackbonetop-1
CLIP400MViT-B/1680.2
ZeroVL100MViT-B/1680.6

Installation

Requirements:

Install requirements:

pip3 install -r requirements.txt

Pre-training

Check PRETRAINING.md for codebase usage.

Model Zoo

ZeroVL 14M weights: Google Drive, Baidu Pan

ZeroVL 100M weights: Google Drive, Baidu Pan

Evaluation

Check EVALUATION.md for codebase usage.

Linear Probing

Check LINEAR.md for codebase usage.

Citing ZeroVL

If you use ZeroVL in your research or wish to refer to the baseline results, please use the following BibTeX entry.

@article{cui2022zerovl,
  title={Contrastive Vision-Language Pre-training with Limited Resources},
  author={Cui, Quan and Zhou, Boyan and Guo, Yu and Yin, Weidong and Wu, Hao and Yoshie, Osamu and Chen, Yubo},
  journal={ECCV},
  year={2022}
}

License

ZeroVL is released under the MIT license. See LICENSE for details.