Mitigating Task and Data Domain Discrepancy for Open- Vocabulary Video Instance Segmentation
Electronics · 12.08.2026 · 15 (16) · 3585
DOI: 10.3390/electronics15163585 ↗Kurz gefasst
Fachartikel in der Zeitschrift Electronics, online erschienen am 12.08.2026. Verfasst von Jiayi Fan und 4 weiteren Autor:innen. Die Arbeit nennt 51 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Department of Science and Technology of Shandong Province, Talent Work Leading Group of the CPC Jinan Municipal Committee.
Abstract (Originalsprache)
Current Open-Vocabulary Video Instance Segmentation (OVVIS) approaches rely on vision–language models (VLMs) for generalization to unseen categories. However, they often fuse segmentation queries with VLM features in a simple and direct manner, overlooking the domain discrepancy between these two tasks. This mismatch limits both the expressiveness of the queries and the zero-shot potential of VLMs. Moreover, most existing models keep the fixed queries learned from closed-set training data even when the test data distribution changes drastically, failing to enhance the language awareness of the model. To address these challenges, we propose Task and Data Domain-Aware Open-Vocabulary Video Instance Segmentation (TD-OVIS), a task-aligned query adaptation framework. Building upon a dual-branch approach, we improve open-vocabulary generalization by initializing queries with language guidance and classifying with test-time tuned text embeddings to mitigate the task and data domain gap, thereby aligning segmentation representations with the VLM embedding space. Extensive experiments on multiple benchmarks show that TD-OVIS generalizes well to unseen categories and achieves strong performance across multiple open-vocabulary VIS datasets.
Förderung
- Department of Science and Technology of Shandong Province
- Talent Work Leading Group of the CPC Jinan Municipal Committee
Zitieren
APA
Fan, J., Zhou, Y., Guo, K., Zhai, R., & Gong, Y. (2026). Mitigating Task and Data Domain Discrepancy for Open- Vocabulary Video Instance Segmentation. Electronics, 15(16), 3585. https://doi.org/10.3390/electronics15163585
Vancouver
Fan J, Zhou Y, Guo K, Zhai R, Gong Y. Mitigating Task and Data Domain Discrepancy for Open- Vocabulary Video Instance Segmentation. Electronics. 2026;15(16):3585. doi:10.3390/electronics15163585
BibTeX
@article{Fan2026,
author = {Fan, Jiayi and Zhou, Yongwang and Guo, Kai and Zhai, Rui and Gong, Yongshun},
title = {Mitigating Task and Data Domain Discrepancy for Open- Vocabulary Video Instance Segmentation},
journal = {Electronics},
year = {2026},
volume = {15},
number = {16},
pages = {3585},
doi = {10.3390/electronics15163585}
}
EUVIMED