Long-Tailed Object Detection Pre-training: Dynamic Rebalancing Contrastive Learning with Dual Reconstruction
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866916480089587712 |
|---|---|
| author | Duan, Chen-Long Li, Yong Wei, Xiu-Shen Zhao, Lin |
| author_facet | Duan, Chen-Long Li, Yong Wei, Xiu-Shen Zhao, Lin |
| contents | Pre-training plays a vital role in various vision tasks, such as object recognition and detection. Commonly used pre-training methods, which typically rely on randomized approaches like uniform or Gaussian distributions to initialize model parameters, often fall short when confronted with long-tailed distributions, especially in detection tasks. This is largely due to extreme data imbalance and the issue of simplicity bias. In this paper, we introduce a novel pre-training framework for object detection, called Dynamic Rebalancing Contrastive Learning with Dual Reconstruction (2DRCL). Our method builds on a Holistic-Local Contrastive Learning mechanism, which aligns pre-training with object detection by capturing both global contextual semantics and detailed local patterns. To tackle the imbalance inherent in long-tailed data, we design a dynamic rebalancing strategy that adjusts the sampling of underrepresented instances throughout the pre-training process, ensuring better representation of tail classes. Moreover, Dual Reconstruction addresses simplicity bias by enforcing a reconstruction task aligned with the self-consistency principle, specifically benefiting underrepresented tail classes. Experiments on COCO and LVIS v1.0 datasets demonstrate the effectiveness of our method, particularly in improving the mAP/AP scores for tail classes. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_09453 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Long-Tailed Object Detection Pre-training: Dynamic Rebalancing Contrastive Learning with Dual Reconstruction Duan, Chen-Long Li, Yong Wei, Xiu-Shen Zhao, Lin Computer Vision and Pattern Recognition Machine Learning Pre-training plays a vital role in various vision tasks, such as object recognition and detection. Commonly used pre-training methods, which typically rely on randomized approaches like uniform or Gaussian distributions to initialize model parameters, often fall short when confronted with long-tailed distributions, especially in detection tasks. This is largely due to extreme data imbalance and the issue of simplicity bias. In this paper, we introduce a novel pre-training framework for object detection, called Dynamic Rebalancing Contrastive Learning with Dual Reconstruction (2DRCL). Our method builds on a Holistic-Local Contrastive Learning mechanism, which aligns pre-training with object detection by capturing both global contextual semantics and detailed local patterns. To tackle the imbalance inherent in long-tailed data, we design a dynamic rebalancing strategy that adjusts the sampling of underrepresented instances throughout the pre-training process, ensuring better representation of tail classes. Moreover, Dual Reconstruction addresses simplicity bias by enforcing a reconstruction task aligned with the self-consistency principle, specifically benefiting underrepresented tail classes. Experiments on COCO and LVIS v1.0 datasets demonstrate the effectiveness of our method, particularly in improving the mAP/AP scores for tail classes. |
| title | Long-Tailed Object Detection Pre-training: Dynamic Rebalancing Contrastive Learning with Dual Reconstruction |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2411.09453 |