Diverse Feature-Level Guidance Adjustments for Unsupervised Domain Adaptative Object Detection
<b>Unsupervised Domain Adaptative Object Detection</b> (UDAOD) aims to alleviate the gap between the source domain and the target domain. Previous methods sought to plainly align global and local features across domains but adapted numerous pooled features and overlooked contextual infor...
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2024-03-01
|
Series: | Applied Sciences |
Subjects: | |
Online Access: | https://www.mdpi.com/2076-3417/14/7/2844 |
_version_ | 1797212888177311744 |
---|---|
author | Yuhe Zhu Chang Liu Yunfei Bai Caiju Wang Chengwei Wei Zhenglin Li Yang Zhou |
author_facet | Yuhe Zhu Chang Liu Yunfei Bai Caiju Wang Chengwei Wei Zhenglin Li Yang Zhou |
author_sort | Yuhe Zhu |
collection | DOAJ |
description | <b>Unsupervised Domain Adaptative Object Detection</b> (UDAOD) aims to alleviate the gap between the source domain and the target domain. Previous methods sought to plainly align global and local features across domains but adapted numerous pooled features and overlooked contextual information, which caused incorrect perceptions of foreground information. To tackle these problems, we propose <b>Diverse Feature-level Guidance Adjustments</b> (DFGAs) for two-stage object detection frameworks, including <b>Pixel-wise Multi-scale Alignment</b> (PMA) and <b>Adaptative Threshold Confidence Adjustment</b> (ATCA). Specifically, PMA adapts features within diverse hierarchical levels to capture sufficient contextual information. Through a customized PMA loss, features from different stages of a network facilitate information interaction across domains. Training with this loss function contributes to the generation of more domain-agnostic features. To better recognize foreground and background samples, ATCA employs adaptative thresholds to divide the foreground and background samples. This strategy flexibly instructs the classifier to perceive the significance of box candidates. Comprehensive experiments are conducted on Cityscapes, Foggy Cityscapes, KITTI, and Sim10k datasets to further demonstrate the superior performance of our method compared to the baseline method. |
first_indexed | 2024-04-24T10:49:32Z |
format | Article |
id | doaj.art-037bb4d89ce74a61bf61c22533e80cf0 |
institution | Directory Open Access Journal |
issn | 2076-3417 |
language | English |
last_indexed | 2024-04-24T10:49:32Z |
publishDate | 2024-03-01 |
publisher | MDPI AG |
record_format | Article |
series | Applied Sciences |
spelling | doaj.art-037bb4d89ce74a61bf61c22533e80cf02024-04-12T13:14:59ZengMDPI AGApplied Sciences2076-34172024-03-01147284410.3390/app14072844Diverse Feature-Level Guidance Adjustments for Unsupervised Domain Adaptative Object DetectionYuhe Zhu0Chang Liu1Yunfei Bai2Caiju Wang3Chengwei Wei4Zhenglin Li5Yang Zhou6Research Institute of USV Engineering, School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, ChinaResearch Institute of USV Engineering, School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, ChinaResearch Institute of USV Engineering, School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, ChinaResearch Institute of USV Engineering, School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, ChinaResearch Institute of USV Engineering, School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, ChinaInstitute of Artifcial Intelligence, Shanghai University, Shanghai 200444, ChinaResearch Institute of USV Engineering, School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, China<b>Unsupervised Domain Adaptative Object Detection</b> (UDAOD) aims to alleviate the gap between the source domain and the target domain. Previous methods sought to plainly align global and local features across domains but adapted numerous pooled features and overlooked contextual information, which caused incorrect perceptions of foreground information. To tackle these problems, we propose <b>Diverse Feature-level Guidance Adjustments</b> (DFGAs) for two-stage object detection frameworks, including <b>Pixel-wise Multi-scale Alignment</b> (PMA) and <b>Adaptative Threshold Confidence Adjustment</b> (ATCA). Specifically, PMA adapts features within diverse hierarchical levels to capture sufficient contextual information. Through a customized PMA loss, features from different stages of a network facilitate information interaction across domains. Training with this loss function contributes to the generation of more domain-agnostic features. To better recognize foreground and background samples, ATCA employs adaptative thresholds to divide the foreground and background samples. This strategy flexibly instructs the classifier to perceive the significance of box candidates. Comprehensive experiments are conducted on Cityscapes, Foggy Cityscapes, KITTI, and Sim10k datasets to further demonstrate the superior performance of our method compared to the baseline method.https://www.mdpi.com/2076-3417/14/7/2844unsupervised domain adaptative object detectionfeature distributionfeature alignmentforeground–background sample division |
spellingShingle | Yuhe Zhu Chang Liu Yunfei Bai Caiju Wang Chengwei Wei Zhenglin Li Yang Zhou Diverse Feature-Level Guidance Adjustments for Unsupervised Domain Adaptative Object Detection Applied Sciences unsupervised domain adaptative object detection feature distribution feature alignment foreground–background sample division |
title | Diverse Feature-Level Guidance Adjustments for Unsupervised Domain Adaptative Object Detection |
title_full | Diverse Feature-Level Guidance Adjustments for Unsupervised Domain Adaptative Object Detection |
title_fullStr | Diverse Feature-Level Guidance Adjustments for Unsupervised Domain Adaptative Object Detection |
title_full_unstemmed | Diverse Feature-Level Guidance Adjustments for Unsupervised Domain Adaptative Object Detection |
title_short | Diverse Feature-Level Guidance Adjustments for Unsupervised Domain Adaptative Object Detection |
title_sort | diverse feature level guidance adjustments for unsupervised domain adaptative object detection |
topic | unsupervised domain adaptative object detection feature distribution feature alignment foreground–background sample division |
url | https://www.mdpi.com/2076-3417/14/7/2844 |
work_keys_str_mv | AT yuhezhu diversefeaturelevelguidanceadjustmentsforunsuperviseddomainadaptativeobjectdetection AT changliu diversefeaturelevelguidanceadjustmentsforunsuperviseddomainadaptativeobjectdetection AT yunfeibai diversefeaturelevelguidanceadjustmentsforunsuperviseddomainadaptativeobjectdetection AT caijuwang diversefeaturelevelguidanceadjustmentsforunsuperviseddomainadaptativeobjectdetection AT chengweiwei diversefeaturelevelguidanceadjustmentsforunsuperviseddomainadaptativeobjectdetection AT zhenglinli diversefeaturelevelguidanceadjustmentsforunsuperviseddomainadaptativeobjectdetection AT yangzhou diversefeaturelevelguidanceadjustmentsforunsuperviseddomainadaptativeobjectdetection |