DRCO: Dense-Label Refinement and Cross Optimization for Semi-Supervised Object Detection
In semi-supervised object detection (SSOD), the methods based on dense pseudo-labeling bypass complex post-processing while maintaining competitive performance compared to the methods based on sparse pseudo-labeling. However, there are still relatively few researches focused on the dense pseudo-labe...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10818482/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841533393680990208 |
|---|---|
| author | Yunlong Qin Yanjun Li Feifan Ji Yan Liu Yu Wang Ji Xiang |
| author_facet | Yunlong Qin Yanjun Li Feifan Ji Yan Liu Yu Wang Ji Xiang |
| author_sort | Yunlong Qin |
| collection | DOAJ |
| description | In semi-supervised object detection (SSOD), the methods based on dense pseudo-labeling bypass complex post-processing while maintaining competitive performance compared to the methods based on sparse pseudo-labeling. However, there are still relatively few researches focused on the dense pseudo-labeling paradigm. In this work, we first experimentally point out the shortcomings of current dense pseudo-labeling methods: 1) Low-quality sampling: the fixed threshold strategies can result in numerous false negatives and false positives. 2) Inconsistency between classification scores and localization quality: classification scores cannot represent localization quality, resulting in poor quality of predicted bounding boxes for sampled positive samples. 3) Suboptimal training approach: current training methods only utilize its knowledge from the perspective of final dense pseudo-labels, failing to fully exploit the teacher model. To address these issues, we propose a method called Dense-Label Refinement and Cross Optimization (DRCO) based on dense pseudo-labels. Specifically, to tackle the issue of low-quality sampling, we introduce the Adaptive Sampling Approach (ASA), which achieves high-quality sampling at the image level and dynamic sampling ratios without introducing any additional hyperparameters. For the inconsistency between classification scores and localization quality, Comprehensive FRS (cFRS) is proposed to jointly optimize the classification and localization branches more efficiently, thereby obtaining a more comprehensive score. Finally, for the suboptimal training approach, we introduce Cross Prediction Optimization (CPO). CPO efficiently leverages the knowledge of the teacher model through a Cross-Head operation, thereby achieving more effective teacher-student interaction. DRCO achieves 27.31% mAP with only 1% COCO labeled data, which is approximately 1.24% mAP higher than the previous state-of-the-art. Based on MS COCO and PASCAL VOC benchmarks, further comprehensive experiments demonstrate that our method alleviates the aforementioned shortcomings and achieves competitive performance. |
| format | Article |
| id | doaj-art-2dfa115f731a43cdbf6d754a7e5950a5 |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-2dfa115f731a43cdbf6d754a7e5950a52025-01-16T00:01:46ZengIEEEIEEE Access2169-35362025-01-01133572358210.1109/ACCESS.2024.352402910818482DRCO: Dense-Label Refinement and Cross Optimization for Semi-Supervised Object DetectionYunlong Qin0https://orcid.org/0009-0004-7822-962XYanjun Li1Feifan Ji2Yan Liu3Yu Wang4https://orcid.org/0000-0001-9066-3294Ji Xiang5https://orcid.org/0000-0002-7234-6460School of Information and Electrical Engineering, Hangzhou City University, Hangzhou, ChinaSchool of Information and Electrical Engineering, Hangzhou City University, Hangzhou, ChinaSchool of Information and Electrical Engineering, Hangzhou City University, Hangzhou, ChinaSchool of Information and Electrical Engineering, Hangzhou City University, Hangzhou, ChinaSchool of Information and Electrical Engineering, Hangzhou City University, Hangzhou, ChinaCollege of Electrical Engineering, Zhejiang University, Hangzhou, ChinaIn semi-supervised object detection (SSOD), the methods based on dense pseudo-labeling bypass complex post-processing while maintaining competitive performance compared to the methods based on sparse pseudo-labeling. However, there are still relatively few researches focused on the dense pseudo-labeling paradigm. In this work, we first experimentally point out the shortcomings of current dense pseudo-labeling methods: 1) Low-quality sampling: the fixed threshold strategies can result in numerous false negatives and false positives. 2) Inconsistency between classification scores and localization quality: classification scores cannot represent localization quality, resulting in poor quality of predicted bounding boxes for sampled positive samples. 3) Suboptimal training approach: current training methods only utilize its knowledge from the perspective of final dense pseudo-labels, failing to fully exploit the teacher model. To address these issues, we propose a method called Dense-Label Refinement and Cross Optimization (DRCO) based on dense pseudo-labels. Specifically, to tackle the issue of low-quality sampling, we introduce the Adaptive Sampling Approach (ASA), which achieves high-quality sampling at the image level and dynamic sampling ratios without introducing any additional hyperparameters. For the inconsistency between classification scores and localization quality, Comprehensive FRS (cFRS) is proposed to jointly optimize the classification and localization branches more efficiently, thereby obtaining a more comprehensive score. Finally, for the suboptimal training approach, we introduce Cross Prediction Optimization (CPO). CPO efficiently leverages the knowledge of the teacher model through a Cross-Head operation, thereby achieving more effective teacher-student interaction. DRCO achieves 27.31% mAP with only 1% COCO labeled data, which is approximately 1.24% mAP higher than the previous state-of-the-art. Based on MS COCO and PASCAL VOC benchmarks, further comprehensive experiments demonstrate that our method alleviates the aforementioned shortcomings and achieves competitive performance.https://ieeexplore.ieee.org/document/10818482/Semi-supervised object detectionobject detectionsemi-supervised learning |
| spellingShingle | Yunlong Qin Yanjun Li Feifan Ji Yan Liu Yu Wang Ji Xiang DRCO: Dense-Label Refinement and Cross Optimization for Semi-Supervised Object Detection IEEE Access Semi-supervised object detection object detection semi-supervised learning |
| title | DRCO: Dense-Label Refinement and Cross Optimization for Semi-Supervised Object Detection |
| title_full | DRCO: Dense-Label Refinement and Cross Optimization for Semi-Supervised Object Detection |
| title_fullStr | DRCO: Dense-Label Refinement and Cross Optimization for Semi-Supervised Object Detection |
| title_full_unstemmed | DRCO: Dense-Label Refinement and Cross Optimization for Semi-Supervised Object Detection |
| title_short | DRCO: Dense-Label Refinement and Cross Optimization for Semi-Supervised Object Detection |
| title_sort | drco dense label refinement and cross optimization for semi supervised object detection |
| topic | Semi-supervised object detection object detection semi-supervised learning |
| url | https://ieeexplore.ieee.org/document/10818482/ |
| work_keys_str_mv | AT yunlongqin drcodenselabelrefinementandcrossoptimizationforsemisupervisedobjectdetection AT yanjunli drcodenselabelrefinementandcrossoptimizationforsemisupervisedobjectdetection AT feifanji drcodenselabelrefinementandcrossoptimizationforsemisupervisedobjectdetection AT yanliu drcodenselabelrefinementandcrossoptimizationforsemisupervisedobjectdetection AT yuwang drcodenselabelrefinementandcrossoptimizationforsemisupervisedobjectdetection AT jixiang drcodenselabelrefinementandcrossoptimizationforsemisupervisedobjectdetection |