Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

CCMB: A Large-scale Chinese Cross-modal Benchmark

About

Vision-language pre-training (VLP) on large-scale datasets has shown premier performance on various downstream tasks. In contrast to plenty of available benchmarks with English corpus, large-scale pre-training datasets and downstream datasets with Chinese corpus remain largely unexplored. In this work, we build a large-scale high-quality Chinese Cross-Modal Benchmark named CCMB for the research community, which contains the currently largest public pre-training dataset Zero and five human-annotated fine-tuning datasets for downstream tasks. Zero contains 250 million images paired with 750 million text descriptions, plus two of the five fine-tuning datasets are also currently the largest ones for Chinese cross-modal downstream tasks. Along with the CCMB, we also develop a VLP framework named R2D2, applying a pre-Ranking + Ranking strategy to learn powerful vision-language representations and a two-way distillation method (i.e., target-guided Distillation and feature-guided Distillation) to further enhance the learning capability. With the Zero and the R2D2 VLP framework, we achieve state-of-the-art performance on twelve downstream datasets from five broad categories of tasks including image-text retrieval, image-text matching, image caption, text-to-image generation, and zero-shot image classification. The datasets, models, and codes are available at https://github.com/yuxie11/R2D2

Chunyu Xie, Heng Cai, Jincheng Li, Fanjing Kong, Xiaoyu Wu, Jianfei Song, Henrique Morimitsu, Lin Yao, Dexin Wang, Xiangzheng Zhang, Dawei Leng, Baochang Zhang, Xiangyang Ji, Yafeng Deng• 2022

Related benchmarks

TaskDatasetResultRank
Text-to-Image RetrievalMSCOCO (1K test)
R@16.33e+3
104
Text-to-Image RetrievalFlickr30K-CN
R@184.4
99
Image-to-Text RetrievalFlickr30K-CN
R@195.6
99
Image-to-Text RetrievalMSCOCO (1K test)
R@156.4
82
Text-to-Image RetrievalCOCO-CN
R@156.4
49
Image-to-Text RetrievalCOCO-CN
R@163.3
48
Text-to-Image RetrievalMUGE
R@160.1
40
Image-to-Text RetrievalAIC-ICC
R@130.7
17
Image-Text RetrievalCBVS-20K (test)
R@141.8
16
Image-to-Text RetrievalFlickr30K-CN 1K (test)
R@195.6
12
Showing 10 of 29 rows

Other info

Code

Follow for update