Author

Zhirong Wu

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Sep 2026

CoViT: Instance-Correspondence Contrastive Learning for Vision Transformer

CoViT consistently achieves stable performance gains of over 2 AP points across multiple instance-level perception tasks by using ViT as backbone architecture, demonstrating that a pure ViT can learn instance-aware representations via inherent attention priors and targeted contrastive constraints.

Yisen Wang, Zhirong Wu, Li-Min Wang · 0 citations