| Cascading Alignment for Unsupervised Domain-Adaptive DETR with Improved DeNoising Anchor Boxes | |
| Geng, Huantong1,2; Jiang, Jun1; Shen, Junye1; Hou, Mengmeng1 | |
| 2022-12 | |
| 发表期刊 | SENSORS
 |
| 卷号 | 22期号:24 |
| 摘要 | Transformer-based object detection has recently attracted increasing interest and shown promising results. As one of the DETR-like models, DETR with improved denoising anchor boxes (DINO) produced superior performance on COCO val2017 and achieved a new state of the art. However, it often encounters challenges when applied to new scenarios where no annotated data is available, and the imaging conditions differ significantly. To alleviate this problem of domain shift, in this paper, unsupervised domain adaptive DINO via cascading alignment (CA-DINO) was proposed, which consists of attention-enhanced double discriminators (AEDD) and weak-restraints on category-level token (WROT). Specifically, AEDD is used to aggregate and align the local-global context from the feature representations of both domains while reducing the domain discrepancy before entering the transformer encoder and decoder. WROT extends Deep CORAL loss to adapt class tokens after embedding, minimizing the difference in second-order statistics between the source and target domain. Our approach is trained end to end, and experiments on two challenging benchmarks demonstrate the effectiveness of our method, which yields 41% relative improvement compared to baseline on the benchmark dataset Foggy Cityscapes, in particular. |
| 关键词 | object detection detection transformer domain adaptation DINO |
| DOI | 10.3390/s22249629 |
| 收录类别 | SCIE |
| 语种 | 英语 |
| WOS研究方向 | Chemistry ; Engineering ; Instruments & Instrumentation |
| WOS类目 | Chemistry, Analytical ; Engineering, Electrical & Electronic ; Instruments & Instrumentation |
| WOS记录号 | WOS:000902819900001 |
| 出版者 | MDPI |
| 原始文献类型 | Article |
| EISSN | 1424-8220 |
| 引用统计 | |
| 文献类型 | 期刊论文 |
| 条目标识符 | http://ir.library.ouchn.edu.cn/handle/39V7QQFX/168577 |
| 专题 | 国家开放大学江苏分部 |
| 通讯作者 | Jiang, Jun |
| 作者单位 | 1.Nanjing Univ Informat Sci & Technol, Sch Comp Sci, Nanjing 210044, Peoples R China; 2.Jiangsu Open Univ, Sch Informat Technol, Nanjing 210036, Peoples R China |
| 第一作者单位 | 国家开放大学江苏分部 |
| 推荐引用方式 GB/T 7714 | Geng, Huantong,Jiang, Jun,Shen, Junye,et al. Cascading Alignment for Unsupervised Domain-Adaptive DETR with Improved DeNoising Anchor Boxes[J]. SENSORS,2022,22(24). |
| APA | Geng, Huantong,Jiang, Jun,Shen, Junye,&Hou, Mengmeng.(2022).Cascading Alignment for Unsupervised Domain-Adaptive DETR with Improved DeNoising Anchor Boxes.SENSORS,22(24). |
| MLA | Geng, Huantong,et al."Cascading Alignment for Unsupervised Domain-Adaptive DETR with Improved DeNoising Anchor Boxes".SENSORS 22.24(2022). |
| 条目包含的文件 | 条目无相关文件。 | |||||
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
修改评论