English

MixGen: A New Multi-Modal Data Augmentation

Computer Vision and Pattern Recognition 2023-01-11 v3 Artificial Intelligence Machine Learning

Abstract

Data augmentation is a necessity to enhance data efficiency in deep learning. For vision-language pre-training, data is only augmented either for images or for text in previous works. In this paper, we present MixGen: a joint data augmentation for vision-language representation learning to further improve data efficiency. It generates new image-text pairs with semantic relationships preserved by interpolating images and concatenating text. It's simple, and can be plug-and-played into existing pipelines. We evaluate MixGen on four architectures, including CLIP, ViLT, ALBEF and TCL, across five downstream vision-language tasks to show its versatility and effectiveness. For example, adding MixGen in ALBEF pre-training leads to absolute performance improvements on downstream tasks: image-text retrieval (+6.2% on COCO fine-tuned and +5.3% on Flicker30K zero-shot), visual grounding (+0.9% on RefCOCO+), visual reasoning (+$0.9% on NLVR2), visual question answering (+0.3% on VQA2.0), and visual entailment (+0.4% on SNLI-VE).

Keywords

Cite

@article{arxiv.2206.08358,
  title  = {MixGen: A New Multi-Modal Data Augmentation},
  author = {Xiaoshuai Hao and Yi Zhu and Srikar Appalaraju and Aston Zhang and Wanqian Zhang and Bo Li and Mu Li},
  journal= {arXiv preprint arXiv:2206.08358},
  year   = {2023}
}

Comments

First three authors contributed equally. Code are available at https://github.com/amazon-research/mix-generation. Oral presentation at WACV 2023 Pretraining Large Vision and Multimodal Models Workshop

R2 v1 2026-06-24T11:54:14.166Z