EUVIMEDEUVIMED entdecken
Crossref

GVLTrack: Global Vision-Language Tracking with Multi-Stage Modal Fusion

Jian Tao, Sixian Chan, Cong Bai, Xiaoqin Zhang

ACM Transactions on Multimedia Computing, Communications, and Applications · 31.05.2026 · 22 (5) · 1-20

DOI: 10.1145/3803863 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ACM Transactions on Multimedia Computing, Communications, and Applications, online erschienen am 21.04.2026. Verfasst von Jian Tao und 3 weiteren Autor:innen. Die Arbeit nennt 61 Literaturangaben. Gefördert von: National Key Research and Development Program Project of China, National Natural Science Foundation of China under, Zhejiang Provincial Natural Science Foundation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In general, local Visual-Language (VL) trackers search targets around the previous bounding box by initial VL annotations. However, there is an inherent contradiction between the local searching perspective of the tracker and the orientation descriptions in language conducted under the global perspective. Furthermore, most methods only fuse modality information in a single stage, which tends to an insufficient relation modeling. To address these issues, we propose a Global Vision-Language Tracker (GVLTrack) with multi-stage modal fusion. First, it tracks the target in the entire image instead of local tracking based on previous results to resolve the above contradiction. Second, GVLTrack incorporates three modal interaction modules: Consistent Relationship Modeling (CRM), VL-Guided Query (VLQ) Initialization, and Recurrent Cross-Modal Decoder (RC-Decoder) to fuse vision-language modality and refine the bounding box progressively comprehensively. We conduct extensive experiments on several benchmarks and achieve competitive performance, demonstrating the effectiveness of our approach. The code will be made publicly available as soon as it is accepted.

Förderung

  • National Key Research and Development Program Project of China
  • National Natural Science Foundation of China under
  • Zhejiang Provincial Natural Science Foundation
  • Fundamental Research Funds for the Provincial Universities of Zhejiang
  • Zhejiang Key Laboratory of Intelligent Manufacturing for Aerodynamic Equipment
  • Hangzhou AI major scientific and technological innovation project

Zitieren

APA

Tao, J., Chan, S., Bai, C., & Zhang, X. (2026). GVLTrack: Global Vision-Language Tracking with Multi-Stage Modal Fusion. ACM Transactions on Multimedia Computing, Communications, and Applications, 22(5), 1-20. https://doi.org/10.1145/3803863

Vancouver

Tao J, Chan S, Bai C, Zhang X. GVLTrack: Global Vision-Language Tracking with Multi-Stage Modal Fusion. ACM Transactions on Multimedia Computing, Communications, and Applications. 2026;22(5):1-20. doi:10.1145/3803863

BibTeX
@article{Tao2026,
  author = {Tao, Jian and Chan, Sixian and Bai, Cong and Zhang, Xiaoqin},
  title = {GVLTrack: Global Vision-Language Tracking with Multi-Stage Modal Fusion},
  journal = {ACM Transactions on Multimedia Computing, Communications, and Applications},
  year = {2026},
  volume = {22},
  number = {5},
  pages = {1--20},
  doi = {10.1145/3803863}
}
Kostenlos starten →