EUVIMEDEUVIMED entdecken
Crossref

Unleashing the Potential of All Test Samples: Mean-Shift Guided Test-Time Adaptation

Jizhou Han, Chenhao Ding, Songlin Dong, Xinyuan Gao, Qiang Wang, Yuhang He, Yihong Gong

IEEE Transactions on Circuits and Systems for Video Technology · 2026-08 · 36 (8) · 11323-11335

DOI: 10.1109/tcsvt.2026.3676383 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Circuits and Systems for Video Technology, 2026-08. Verfasst von Jizhou Han und 6 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 75 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Shenzhen Key Technical Projects, China Postdoctoral Science Foundation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Visual-language models (VLMs) like CLIP exhibit strong generalization but struggle with distribution shifts at test time. Existing training-free test-time adaptation (TTA) methods operate strictly within CLIP’s original feature space, relying on high-confidence samples while overlooking the potential of low-confidence ones. We propose MS-TTA, a training-free approach that enhances feature representations beyond CLIP’s space using a single-step k-nearest neighbors (kNN) Mean-Shift. By refining all test samples, MS-TTA improves feature compactness and class separability, leading to more stable adaptation. Additionally, a cache of refined embeddings further enhances inference by providing Mean-Shift-enhanced logits. Extensive evaluations on OOD and Cross-Dataset Benchmarks demonstrate that MS-TTA consistently outperforms state-of-the-art training-free TTA methods, achieving robust adaptation without requiring additional training.

Förderung

  • National Natural Science Foundation of China
  • Shenzhen Key Technical Projects
  • China Postdoctoral Science Foundation
  • Guangdong Provincial Key Laboratory of Computility Microelectronics

Zitieren

APA

Han, J., Ding, C., Dong, S., Gao, X., Wang, Q., He, Y., & Gong, Y. (2026). Unleashing the Potential of All Test Samples: Mean-Shift Guided Test-Time Adaptation. IEEE Transactions on Circuits and Systems for Video Technology, 36(8), 11323-11335. https://doi.org/10.1109/tcsvt.2026.3676383

Vancouver

Han J, Ding C, Dong S, Gao X, Wang Q, He Y, et al. Unleashing the Potential of All Test Samples: Mean-Shift Guided Test-Time Adaptation. IEEE Transactions on Circuits and Systems for Video Technology. 2026;36(8):11323-11335. doi:10.1109/tcsvt.2026.3676383

BibTeX
@article{Han2026,
  author = {Han, Jizhou and Ding, Chenhao and Dong, Songlin and Gao, Xinyuan and Wang, Qiang and He, Yuhang and Gong, Yihong},
  title = {Unleashing the Potential of All Test Samples: Mean-Shift Guided Test-Time Adaptation},
  journal = {IEEE Transactions on Circuits and Systems for Video Technology},
  year = {2026},
  volume = {36},
  number = {8},
  pages = {11323--11335},
  doi = {10.1109/tcsvt.2026.3676383}
}
Kostenlos starten →