SMCL: Saliency Masked Contrastive Learning for Long-tailed Recognition
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917684486078464 |
|---|---|
| author | Park, Sanglee Hwang, Seung-won So, Jungmin |
| author_facet | Park, Sanglee Hwang, Seung-won So, Jungmin |
| contents | Real-world data often follow a long-tailed distribution with a high imbalance in the number of samples between classes. The problem with training from imbalanced data is that some background features, common to all classes, can be unobserved in classes with scarce samples. As a result, this background correlates to biased predictions into ``major" classes. In this paper, we propose saliency masked contrastive learning, a new method that uses saliency masking and contrastive learning to mitigate the problem and improve the generalizability of a model. Our key idea is to mask the important part of an image using saliency detection and use contrastive learning to move the masked image towards minor classes in the feature space, so that background features present in the masked image are no longer correlated with the original class. Experiment results show that our method achieves state-of-the-art level performance on benchmark long-tailed datasets. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_02223 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | SMCL: Saliency Masked Contrastive Learning for Long-tailed Recognition Park, Sanglee Hwang, Seung-won So, Jungmin Computer Vision and Pattern Recognition Machine Learning Real-world data often follow a long-tailed distribution with a high imbalance in the number of samples between classes. The problem with training from imbalanced data is that some background features, common to all classes, can be unobserved in classes with scarce samples. As a result, this background correlates to biased predictions into ``major" classes. In this paper, we propose saliency masked contrastive learning, a new method that uses saliency masking and contrastive learning to mitigate the problem and improve the generalizability of a model. Our key idea is to mask the important part of an image using saliency detection and use contrastive learning to move the masked image towards minor classes in the feature space, so that background features present in the masked image are no longer correlated with the original class. Experiment results show that our method achieves state-of-the-art level performance on benchmark long-tailed datasets. |
| title | SMCL: Saliency Masked Contrastive Learning for Long-tailed Recognition |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2406.02223 |