Improving Visual Prompt Tuning by Gaussian Neighborhood Minimization for Long-Tailed Visual Recognition

GNM-PT

Temporary demo for GNM-PT.

The core code is in gnm.py.

You May Find Our Additional Works of Interest

• [CVPR'22] Long-Tailed Visual Recognition via Gaussian Clouded Logit Adjustment [paper] [code]

• [TPAMI'23] Key Point Sensitive Loss for Long-Tailed Visual Recognition [paper] [code]

• [CVPR'23] Long-Tailed Visual Recognition via Self-heterogeneous Integration with Knowledge Excavation [paper] [code]

• [AAAI'24] Feature Fusion from Head to Tail for Long-Tailed Visual Recognition [paper] [code]

• [TAI'24] Adjusting logit in Gaussian form for long-tailed visual recognition [paper] [code]

Misc

If you find our paper and repo useful, please cite our paper:

@inproceedings{LiGNMPT,
 author = {Li, Mengke and Liu, Ye and Lu, Yang and Zhang, Yiqun and Cheung, Yiu-ming and Huang, Hui},
 booktitle = {Advances in Neural Information Processing Systems},
 pages = {103985--104009},
 title = {Improving Visual Prompt Tuning by Gaussian Neighborhood Minimization for Long-Tailed Visual Recognition},
 volume = {37},
 year = {2024}
}

Acknowledgment

We refer to the code architecture from VPT reproduce. Many thanks to the authors.