SAM3Dual: A 3rd Place Solution to the MOSEv2 Track, 8th LSVOS Challenge

JeongRae Kim; Chaehyun Kim; Changwon Lim

cs.CVAug 23, 2026

SAM3Dual: A 3rd Place Solution to the MOSEv2 Track, 8th LSVOS Challenge

Authors: JeongRae Kim, Chaehyun Kim, Changwon Lim

Organizations: Chung-Ang University, Seoul, Republic of Korea

Abstract

We present SAM3Dual, our third-place solution to the MOSEv2 track of the 8th Large-scale Video Object Segmentation (LSVOS) Challenge at ECCV 2026. SAM3Dual is a training-free inference extension of pretrained SAM 3 that explicitly separates temporal memory into a short-term branch for recent observations and a long-term branch for interval-sampled historical representations. The two memory responses are combined using a deterministic sequence-relative fusion schedule and conservatively modulated by the previous-frame object confidence. All pretrained SAM 3 parameters remain frozen, requiring no task-specific training, fine-tuning, test-time training, or online parameter optimization. The complete system achieved an official J&F score of 64.37 and ranked third in the MOSEv2 track. This result highlights the potential of reorganizing temporal memory entirely at inference time to obtain competitive long-term VOS performance while preserving the pretrained model.

Figures & tables

Figure 1: Overview of SAM3Dual. Current-frame features independently attend to short-term memory containing recent observations and long-term memory containing interval-sampled historical observations. Their responses are confidence-modulated and combined using a sequence-relative temporal fusion schedule before being passed to the frozen SAM 3 prediction pathway.

Setting	Value
Base model	facebook/sam3
Hardware	1 $\times$ NVIDIA GeForce RTX 4090
Precision	BF16 mixed precision
Training / adaptation	None
Short-term memory	$1+6$ entries
Long-term memory	$1+6$ entries

Table 1: Inference configuration of the official SAM3Dual submission.

Rank	Team / Account	$\boldsymbol{\mathcal{J}\&\mathcal{F}}\uparrow$
1	HITsz-Dragon	69.82
2	mmm	66.20
3	AISTAT ( kjeong )	64.37

Table 2: Top-three entries on the official MOSEv2 leaderboard of the 8th LSVOS Challenge at ECCV 2026.

Figure 2: Representative successful cases of SAM3Dual on the MOSEv2 training set. Green, red, and blue indicate true positives, false positives, and false negatives, respectively.

Figure 3: Representative failure cases on the MOSEv2 training set, primarily involving visually similar distractors, large target displacement, and reappearance. Colors follow Fig. 2 .

Video Object Segmentation

Video object segmentation assigns pixel-level masks to objects across frames while preserving their identities through motion, occlusion, and change.

See all 110 papers

SAM2Dual: Training-Free, Dual Memory for Long-Term Video Object Segmentation
Aug 19, 2026JeongRae Kim, Changwon LimVideo Object Segmentation Memory-Augmented Video Understanding
Department of Applied Statistics, Chung-Ang University, Seoul 06974, Republic of Korea
Competitive Memory Readout for Robust Video Object Segmentation: 2nd Place Technical Report for the MOSEv2 Track of the 8th LSVOS Challenge
Aug 22, 2026Mingqi Gao, Sijie Li, Jungong HanVideo Object Segmentation Segment Anything Models
School of Computer Science, University of Sheffield · Department of Automation, Tsinghua University
VOS-Agent: The 1st Place Solution for the 8th LSVOS Challenge (MOSEv2 Track)
Aug 13, 2026Canyang Wu, Jinrong Zhang, Xusheng He +3Video Object Segmentation Visual Object Tracking
Harbin Institute of Technology, Shenzhen · Nanyang Technological University · Shenzhen Loop Area Institute

Aug 19, 2026J/K move · Enter open · S save

SAM2Dual: Training-Free, Dual Memory for Long-Term Video Object Segmentation

JeongRae Kim, Changwon Lim

Department of Applied Statistics, Chung-Ang University, Seoul 06974, Republic of Korea

Video Object Segmentation Memory-Augmented Video Understanding

Open paper AskPDF arXiv

Long-term video object segmentation (VOS) remains challenging due to error accumulation under extended occlusions, re-appearance, and scene changes. Although SAM2 provides strong zero-shot performance, its streaming memory can amplify drift over long horizons when recent, unreliable predictions dominate the memory state. We propose SAM2Dual, a training-free, plug-and-play inference-time enhancement that improves long-video robustness without updating model weights. SAM2Dual introduces a Dual Memory design that explicitly separates (i) short-term memory for rapid local adaptation and (ii) long-term memory built via interval-based sampling to preserve global identity cues, combined through a gated fusion strategy. In addition, we present Text-Aware Memory (TAM), which extracts a compact word-level cue from early frames and uses text embeddings to reweight memory contributions based on semantic compatibility, supporting identity preservation when visual evidence becomes weak or ambiguous. Across long-term benchmarks, SAM2Dual consistently improves stability on long videos, raising J&F from 49.33 to 50.65 on MOSEv2 and achieving consistent gains on LVOSv2.

Abstract

Video Object Segmentation

Explore similar work

SAM2Dual: Training-Free, Dual Memory for Long-Term Video Object Segmentation

Competitive Memory Readout for Robust Video Object Segmentation: 2nd Place Technical Report for the MOSEv2 Track of the 8th LSVOS Challenge

VOS-Agent: The 1st Place Solution for the 8th LSVOS Challenge (MOSEv2 Track)