Domain Diversification Progressive Cross-Domain Weakly-Supervised Real-time Object Detection

Aiming at the problems of large image translation bias at the pixel-level adaptation, the risk of source-bias discrimination at the feature-level adaptation, and the inability of weakly supervised learning to balance detection accuracy and real- time performance, a diversified domain shifter and pse...

Full description

Saved in:
Bibliographic Details
Main Authors: LI Chengyan, ZHENG Qisen, WANG Hao
Format: Article
Language:zho
Published: Harbin University of Science and Technology Publications 2024-06-01
Series:Journal of Harbin University of Science and Technology
Subjects:
Online Access:https://hlgxb.hrbust.edu.cn/#/digest?ArticleID=2326
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849702325449392128
author LI Chengyan
ZHENG Qisen
WANG Hao
author_facet LI Chengyan
ZHENG Qisen
WANG Hao
author_sort LI Chengyan
collection DOAJ
description Aiming at the problems of large image translation bias at the pixel-level adaptation, the risk of source-bias discrimination at the feature-level adaptation, and the inability of weakly supervised learning to balance detection accuracy and real- time performance, a diversified domain shifter and pseudo bounding box generator are proposed to gradually adjust the pre-training model. The adaptive cross-domain framework is gradually completed at pixel-level and feature-level. A diversified intermediate domain adjustment detection model is generated from the source domain by a domain shifter to bridge the domain gap and reduce the image translation bias. The intermediate domain is used as the supervised source domain, and the pseudo-labeled image adjustment detection model is generated by combining image-level annotations in the target domain to improve source-bias discrimination. A real-time object detector matching the cross-domain framework is constructed based on SSD algorithm to realize real-time object detection under weakly supervised conditions. The mAP on PASCAL VOC migrated to Clipart1k and other datasets is 0. 4% ~ 4. 7% better than the existing methods. The detection speed is 32 FPS ~47 FPS. This improves the accuracy and meets the requirements of real-time detection, and has better migration detection performance.
format Article
id doaj-art-9f75a6ddbe9c4a05a1b79d7fda460e01
institution DOAJ
issn 1007-2683
language zho
publishDate 2024-06-01
publisher Harbin University of Science and Technology Publications
record_format Article
series Journal of Harbin University of Science and Technology
spelling doaj-art-9f75a6ddbe9c4a05a1b79d7fda460e012025-08-20T03:17:40ZzhoHarbin University of Science and Technology PublicationsJournal of Harbin University of Science and Technology1007-26832024-06-012903111910.15938/j.jhust.2024.03.002Domain Diversification Progressive Cross-Domain Weakly-Supervised Real-time Object DetectionLI Chengyan0ZHENG Qisen1WANG Hao2School of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080,ChinaSchool of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080,ChinaSchool of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080,ChinaAiming at the problems of large image translation bias at the pixel-level adaptation, the risk of source-bias discrimination at the feature-level adaptation, and the inability of weakly supervised learning to balance detection accuracy and real- time performance, a diversified domain shifter and pseudo bounding box generator are proposed to gradually adjust the pre-training model. The adaptive cross-domain framework is gradually completed at pixel-level and feature-level. A diversified intermediate domain adjustment detection model is generated from the source domain by a domain shifter to bridge the domain gap and reduce the image translation bias. The intermediate domain is used as the supervised source domain, and the pseudo-labeled image adjustment detection model is generated by combining image-level annotations in the target domain to improve source-bias discrimination. A real-time object detector matching the cross-domain framework is constructed based on SSD algorithm to realize real-time object detection under weakly supervised conditions. The mAP on PASCAL VOC migrated to Clipart1k and other datasets is 0. 4% ~ 4. 7% better than the existing methods. The detection speed is 32 FPS ~47 FPS. This improves the accuracy and meets the requirements of real-time detection, and has better migration detection performance.https://hlgxb.hrbust.edu.cn/#/digest?ArticleID=2326real-time object detectionweakly supervised learningdomain adaptationimage translation networkssd algorithm
spellingShingle LI Chengyan
ZHENG Qisen
WANG Hao
Domain Diversification Progressive Cross-Domain Weakly-Supervised Real-time Object Detection
Journal of Harbin University of Science and Technology
real-time object detection
weakly supervised learning
domain adaptation
image translation network
ssd algorithm
title Domain Diversification Progressive Cross-Domain Weakly-Supervised Real-time Object Detection
title_full Domain Diversification Progressive Cross-Domain Weakly-Supervised Real-time Object Detection
title_fullStr Domain Diversification Progressive Cross-Domain Weakly-Supervised Real-time Object Detection
title_full_unstemmed Domain Diversification Progressive Cross-Domain Weakly-Supervised Real-time Object Detection
title_short Domain Diversification Progressive Cross-Domain Weakly-Supervised Real-time Object Detection
title_sort domain diversification progressive cross domain weakly supervised real time object detection
topic real-time object detection
weakly supervised learning
domain adaptation
image translation network
ssd algorithm
url https://hlgxb.hrbust.edu.cn/#/digest?ArticleID=2326
work_keys_str_mv AT lichengyan domaindiversificationprogressivecrossdomainweaklysupervisedrealtimeobjectdetection
AT zhengqisen domaindiversificationprogressivecrossdomainweaklysupervisedrealtimeobjectdetection
AT wanghao domaindiversificationprogressivecrossdomainweaklysupervisedrealtimeobjectdetection