《计算机应用》唯一官方网站 ›› 2025, Vol. 45 ›› Issue (8): 2546-2554.DOI: 10.11772/j.issn.1001-9081.2024081096

• 人工智能 • 上一篇    

自步学习指导下的半监督目标检测框架

谢斌红, 剌颖坤(), 张英俊, 张睿   

  1. 太原科技大学 计算机科学与技术学院,太原 030024
  • 收稿日期:2024-08-06 修回日期:2024-10-21 接受日期:2024-10-28 发布日期:2024-11-19 出版日期:2025-08-10
  • 通讯作者: 剌颖坤
  • 作者简介:谢斌红(1971—),男,山西万荣人,教授,硕士,主要研究方向:智能化软件、机器学习
    张英俊(1969—),男,山西河津人,教授级高级工程师,硕士,主要研究方向:智能化软件、软件体系结构
    张睿(1987—),男,山西太原人,副教授,博士,主要研究方向:智能信息处理。
  • 基金资助:
    山西省基础研究计划项目(面上)(20210302123216);吕梁市引进高层次科技人才重点研发项目(2022RC08)

Semi-supervised object detection framework guided by self-paced learning

Binhong XIE, Yingkun LA(), Yingjun ZHANG, Rui ZHANG   

  1. College of Computer Science and Technology,Taiyuan University of Science and Technology,Taiyuan Shanxi 030024,China
  • Received:2024-08-06 Revised:2024-10-21 Accepted:2024-10-28 Online:2024-11-19 Published:2025-08-10
  • Contact: Yingkun LA
  • About author:XIE Binhong, born in 1971, M. S., professor. His research interests include intelligent software, machine learning.
    ZHANG Yingjun, born in 1969, M. S., professor-level senior engineer. His research interests include intelligent software, software architecture.
    ZHANG Rui, born in 1987, Ph. D., associate professor. His research interests include intelligent information processing.
  • Supported by:
    Shanxi Provincial Basic Research Program (General Program)(20210302123216);Lvliang City Key Research and Development Project for Introduction of High-level Scientific and Technological Talents(2022RC08)

摘要:

为了提高伪标签质量并解决半监督目标检测(SSOD)中的确认偏差问题,提出一种基于动态参数的自步学习(SPL)指导下的SSOD框架。在该框架中设计动态自步参数和连续权重变量,以优化SSOD的效果。具体地,动态自步参数根据模型在训练过程中的实时表现评估样本的难易程度,而连续权重变量则通过比较样本损失与动态自步参数的关系精确评估每个样本在训练中的重要性和可靠性,并对样本中的每个物体都进行精细化权重设计。此外,该框架采用单一模型迭代训练,并引入一致性正则化策略评估模型预测的一致性。这种设计不仅能为模型提供更有针对性的权重信息,还可使模型通过权重信息的动态调整自适应地优化训练过程。在PASCAL VOC和MS-COCO数据集上的广泛对比实验结果不仅显示所提框架显著提升了模型的检测精度,还验证了所提框架良好的通用性和高效的收敛性能。特别在PASCAL VOC数据集上,所提框架相较于LabelMatch、Unbiased Teacher V2和MixTeacher在检测精度上分别提升了0.65、4.84和0.28个百分点。

关键词: 半监督目标检测, 自步学习, 一致性正则化, 动态自步参数, 连续权重变量

Abstract:

In order to improve the quality of pseudo-labels and solve the problem of confirmation bias in Semi-Supervised Object Detection (SSOD), an SSOD framework based on dynamic parameters under guidance of Self-Paced Learning (SPL) was proposed. In the framework, dynamic self-paced parameter and continuous weight variable were designed to optimize the effect of SSOD. In specific, the dynamic self-paced parameter was used to evaluate difficulty of the samples based on real-time performance of the model during training process, the continuous weight variable was used to evaluate importance and reliability of each sample in training accurately by comparing relationship between sample loss and dynamic self-paced parameters, and refine weight design of each object in the samples. In addition, a single model was used in the framework for iterative training, and a consistency regularization strategy was introduced to evaluate consistency of the model predictions. This design provided more targeted weight information for the model, and optimized the training process adaptively by the model through dynamic adjustment of the weight information. Extensive comparison experimental results on PASCAL VOC and MS-COCO datasets show that the proposed framework improves the detection accuracy of the model significantly, and verify good generality and efficient convergence performance of the framework. Especially on PASCAL VOC dataset, the proposed framework has the detection precision improved by 0.65, 4.84, and 0.28 percentage points, respectively, compared with LabelMatch, Unbiased Teacher V2, and MixTeacher.

Key words: semi-supervised object detection, Self-Paced Learning (SPL), consistency regularization, dynamic self-paced parameter, continuous weight variable

中图分类号: