The 1st Solution for MOSEv2 Challenge 2025: Long-term and Concept-aware Video Segmentation via SeC

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gao, Mingqi, Chen, Jingkun, Miao, Yunqi, Wu, Gengshen, Qin, Zhijin, Han, Jungong
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911172194729984
author Gao, Mingqi
Chen, Jingkun
Miao, Yunqi
Wu, Gengshen
Qin, Zhijin
Han, Jungong
author_facet Gao, Mingqi
Chen, Jingkun
Miao, Yunqi
Wu, Gengshen
Qin, Zhijin
Han, Jungong
contents This technical report explores the MOSEv2 track of the LSVOS Challenge, which targets complex semi-supervised video object segmentation. By analysing and adapting SeC, an enhanced SAM-2 framework, we conduct a detailed study of its long-term memory and concept-aware memory, showing that long-term memory preserves temporal continuity under occlusion and reappearance, while concept-aware memory supplies semantic priors that suppress distractors; together, these traits directly benefit several MOSEv2's core challenges. Our solution achieves a JF score of 39.89% on the test set, ranking 1st in the MOSEv2 track of the LSVOS Challenge.
format Preprint
id arxiv_https___arxiv_org_abs_2509_19183
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle The 1st Solution for MOSEv2 Challenge 2025: Long-term and Concept-aware Video Segmentation via SeC
Gao, Mingqi
Chen, Jingkun
Miao, Yunqi
Wu, Gengshen
Qin, Zhijin
Han, Jungong
Computer Vision and Pattern Recognition
This technical report explores the MOSEv2 track of the LSVOS Challenge, which targets complex semi-supervised video object segmentation. By analysing and adapting SeC, an enhanced SAM-2 framework, we conduct a detailed study of its long-term memory and concept-aware memory, showing that long-term memory preserves temporal continuity under occlusion and reappearance, while concept-aware memory supplies semantic priors that suppress distractors; together, these traits directly benefit several MOSEv2's core challenges. Our solution achieves a JF score of 39.89% on the test set, ranking 1st in the MOSEv2 track of the LSVOS Challenge.
title The 1st Solution for MOSEv2 Challenge 2025: Long-term and Concept-aware Video Segmentation via SeC
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2509.19183