EUVIMEDEUVIMED entdecken
Crossref

Training-Free Spatio-temporal Decoupled Reasoning Video Segmentation with Adaptive Object Memory

Zhengtong Zhu, Jiaqing Fan, Zhixuan Liu, Fanzhang Li

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (16) · 14022-14030

DOI: 10.1609/aaai.v40i16.38413 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Zhengtong Zhu und 3 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Reasoning Video Object Segmentation (ReasonVOS) is a challenging task that requires stable object segmentation across video sequences using implicit and complex textual inputs. Previous methods fine-tune Multimodal Large Language Models (MLLMs) to produce segmentation outputs, which demand substantial resources. Additionally, some existing methods are coupled in the processing of spatio-temporal information, which affects the temporal stability of the model to some extent. To address these issues, we propose Training-Free Spatio-temporal Decoupled Reasoning Video Segmentation with Adaptive Object Memory (SDAM). We aim to design a training-free reasoning video segmentation framework that outperforms existing methods requiring fine-tuning, using only pre-trained models. Meanwhile, we propose an Adaptive Object Memory module that selects and memorizes key objects based on motion cues in different video sequences. Finally, we propose Spatio-temporal Decoupling for stable temporal propagation. In the spatial domain, we achieve precise localization and segmentation of target objects, while in the temporal domain, we leverage key object temporal information to drive stable cross-frame propagation. Our method achieves excellent results on five benchmark datasets, including Ref-YouTubeVOS, Ref-DAVIS17, MeViS, ReasonVOS, and ReVOS.

Zitieren

APA

Zhu, Z., Fan, J., Liu, Z., & Li, F. (2026). Training-Free Spatio-temporal Decoupled Reasoning Video Segmentation with Adaptive Object Memory. Proceedings of the AAAI Conference on Artificial Intelligence, 40(16), 14022-14030. https://doi.org/10.1609/aaai.v40i16.38413

Vancouver

Zhu Z, Fan J, Liu Z, Li F. Training-Free Spatio-temporal Decoupled Reasoning Video Segmentation with Adaptive Object Memory. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(16):14022-14030. doi:10.1609/aaai.v40i16.38413

BibTeX
@article{Zhu2026,
  author = {Zhu, Zhengtong and Fan, Jiaqing and Liu, Zhixuan and Li, Fanzhang},
  title = {Training-Free Spatio-temporal Decoupled Reasoning Video Segmentation with Adaptive Object Memory},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {16},
  pages = {14022--14030},
  doi = {10.1609/aaai.v40i16.38413}
}
Kostenlos starten →