Multi-Source Soft Pseudo-Label Learning with Domain Similarity-based Weighting for Semantic Segmentation

被引:2
|
作者
Matsuzaki, Shigemichi [1 ]
Masuzawa, Hiroaki [1 ]
Miura, Jun [1 ]
机构
[1] Toyohashi Univ Technol, Dept Comp Sci & Engn, Hibarigaoka 1-1,Tenpaku Cho, Toyohashi, Aichi, Japan
关键词
D O I
10.1109/IROS55552.2023.10342159
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper describes a method of domain adaptive training for semantic segmentation using multiple source datasets that are not necessarily relevant to the target dataset. We propose a soft pseudo-label generation method by integrating predicted object probabilities from multiple source models. The prediction of each source model is weighted based on the estimated domain similarity between the source and the target datasets to emphasize contribution of a model trained on a source that is more similar to the target and generate reasonable pseudo-labels. We also propose a training method using the soft pseudo-labels considering their entropy to fully exploit information from the source datasets while suppressing the influence of possibly misclassified pixels. The experiments show comparative or better performance than our previous work and another existing multi-source domain adaptation method, and applicability to a variety of target environments.
引用
收藏
页码:5852 / 5857
页数:6
相关论文
共 50 条
  • [21] Multi-source adversarial transfer learning for ultrasound image segmentation with limited similarity
    Zhang, Yifu
    Li, Hongru
    Yang, Tao
    Tao, Rui
    Liu, Zhengyuan
    Shi, Shimeng
    Zhang, Jiansong
    Ma, Ning
    Feng, Wujin
    Zhang, Zhanhu
    Zhang, Xinyu
    APPLIED SOFT COMPUTING, 2023, 146
  • [22] Multi-Perspective Pseudo-Label Generation and Confidence-Weighted Training for Semi-Supervised Semantic Segmentation
    Hu, Kai
    Chen, Xiaobo
    Chen, Zhineng
    Zhang, Yuan
    Gao, Xieping
    IEEE TRANSACTIONS ON MULTIMEDIA, 2025, 27 : 300 - 311
  • [23] Multi-source transfer learning based on label shared subspace
    Fang, Min
    Guo, Yong
    Zhang, Xiaosong
    Li, Xiao
    PATTERN RECOGNITION LETTERS, 2015, 51 : 101 - 106
  • [24] Multi-source to multi-target domain adaptation method based on similarity measurement
    Wu, Lan
    Wang, Han
    Yao, Yuan
    IET IMAGE PROCESSING, 2024, 18 (01) : 34 - 46
  • [25] SCoAMPS: Semi-Supervised Graph Contrastive Learning Based on Associative Memory Network and Pseudo-Label Similarity
    Gong, Zaigang
    Chen, Siyu
    Dai, Qiangsheng
    Feng, Ying
    Wang, Jiawei
    Zhang, Jinghui
    BIG DATA MINING AND ANALYTICS, 2025, 8 (02): : 273 - 291
  • [26] Semantic-Aware Adaptive Prompt Learning for Universal Multi-Source Domain Adaptation
    Yang, Yuxiang
    Hou, Yun
    Wen, Lu
    Zeng, Pinxian
    Wang, Yan
    IEEE SIGNAL PROCESSING LETTERS, 2024, 31 : 1444 - 1448
  • [27] Multi-modal contrastive mutual learning and pseudo-label re-learning for semi-supervised medical image segmentation
    Zhang, Shuo
    Zhang, Jiaojiao
    Tian, Biao
    Lukasiewicz, Thomas
    Xu, Zhenghua
    MEDICAL IMAGE ANALYSIS, 2023, 83
  • [28] Deeply Hybrid Contrastive Learning Based on Semantic Pseudo-Label for Salient Object Detection in Optical Remote Sensing Images
    Qiu, Yu
    Sun, Yuhang
    Mei, Jie
    Xu, Jing
    IEEE TRANSACTIONS ON MULTIMEDIA, 2024, 26 : 10892 - 10907
  • [29] Domain and Content Adaptive Convolution Based Multi-Source Domain Generalization for Medical Image Segmentation
    Hu, Shishuai
    Liao, Zehui
    Zhang, Jianpeng
    Xia, Yong
    IEEE TRANSACTIONS ON MEDICAL IMAGING, 2023, 42 (01) : 233 - 244
  • [30] Dual uncertainty-guided multi-model pseudo-label learning for semi-supervised medical image segmentation
    Qiu Z.
    Gan W.
    Yang Z.
    Zhou R.
    Gan H.
    Mathematical Biosciences and Engineering, 2024, 21 (02) : 2212 - 2232