English

SiRi: A Simple Selective Retraining Mechanism for Transformer-based Visual Grounding

Computer Vision and Pattern Recognition 2022-07-28 v1

Abstract

In this paper, we investigate how to achieve better visual grounding with modern vision-language transformers, and propose a simple yet powerful Selective Retraining (SiRi) mechanism for this challenging task. Particularly, SiRi conveys a significant principle to the research of visual grounding, i.e., a better initialized vision-language encoder would help the model converge to a better local minimum, advancing the performance accordingly. In specific, we continually update the parameters of the encoder as the training goes on, while periodically re-initialize rest of the parameters to compel the model to be better optimized based on an enhanced encoder. SiRi can significantly outperform previous approaches on three popular benchmarks. Specifically, our method achieves 83.04% Top1 accuracy on RefCOCO+ testA, outperforming the state-of-the-art approaches (training from scratch) by more than 10.21%. Additionally, we reveal that SiRi performs surprisingly superior even with limited training data. We also extend it to transformer-based visual grounding models and other vision-language tasks to verify the validity.

Keywords

Cite

@article{arxiv.2207.13325,
  title  = {SiRi: A Simple Selective Retraining Mechanism for Transformer-based Visual Grounding},
  author = {Mengxue Qu and Yu Wu and Wu Liu and Qiqi Gong and Xiaodan Liang and Olga Russakovsky and Yao Zhao and Yunchao Wei},
  journal= {arXiv preprint arXiv:2207.13325},
  year   = {2022}
}

Comments

21 pages (including Supplementary Materials); Accepted to ECCV 2022

R2 v1 2026-06-25T01:15:53.395Z