Pith. sign in

REVIEW 3 major objections 4 minor 2 cited by

Referring Video Object Segmentation via Language-aligned Track Selection

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Frozen SAM2 tokens pick the right track, topping MeViS with 48.6 J&F.

desk verdict Genuinely new way to reuse SAM2 tokens for RVOS, with a credible SOTA on MeViS—but the proposal-coverage assumption needs an oracle check before I'd fully trust the headline number. read the letter →

arxiv 2412.01136 v2 pith:PXA2AY2R submitted 2024-12-02 cs.CV

Seongchan Kim , Woojeong Jin , Sangbeom Lim , Heeji Yoon , Hyunwook Choi , Seungryong Kim This is my paper · ORCID
classification cs.CV
keywords referringvideoobjectsegmentationSAM2tokenslanguagealignmenttrackselectionpseudo-labelingzero-shotgeneralization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

SOLA proposes that referring video object segmentation can be reframed as selecting, among candidate object tracks produced by a frozen SAM2, the one that best matches a natural language expression. The paper claims that SAM2's object tokens—the per-frame object pointers the model already outputs—are compact video-level representations that encode both motion and identity, so a lightweight language-aligned selection module can align them with text features. On MeViS this selection approach reaches J&F 48.6, surpassing the previous best result with only 32.9M trainable parameters, and it transfers zero-shot to Ref-YouTube-VOS and Ref-DA VIS. The core message is that a frozen video segmentation model's internal tokens are a sufficient basis for language-conditioned video object selection, provided the candidate tracks cover the objects a sentence might refer to.

What carries the argument

The load-bearing object is the SAM2 object token: the object pointer produced by SAM2's mask decoder at each frame, concatenated over time to form a video-level token per candidate track. A short-term motion encoder (a 1D convolution along the temporal dimension) refines the token, and a stack of attention layers—inter-object attention across tracks in a frame, motion attention along each track's time axis, and object-to-language cross-attention with text tokens—aligns the visual and linguistic representations. The alignment score is computed by aggregating per-token text similarity, and training uses binary cross-entropy on IoU-based pseudo-labels plus a contrastive alignment loss that pulls matched tokens toward the text anchor and pushes mismatched tokens toward learnable negative anchors.

What would settle it

A concrete test: on a held-out set of videos, restrict the prompt set to a subset of object categories, then ask whether SOLA can still select the referred track for a held-out category. If the selection accuracy drops to zero whenever the candidate set excludes the referent, the claim that selection rather than proposal carries the performance is falsified. More quantitatively, compute the recall of candidate tracks with mIoU above 0.5 against ground-truth tracks on MeViS; if that recall is substantially below the reported J&F, then proposal-stage failures, not language alignment, dominate the error.

Watch

Extended reading notes

Core claim

The central claim is that SAM2's object token representations, extracted along with mask tracks, contain enough spatio-temporal objectness that a lightweight module can align them with referring expressions and select the correct track. SOLA operationalizes this by precomputing candidate tracks with fully frozen SAM2, representing each track as a sequence of object pointers over time, and training a small selection network with IoU-based pseudo-labels derived from ground-truth mask tracks. The selection module outputs an alignment score per track, and tracks scoring above a threshold are merged into the final segmentation. The result is a new state of the art on MeViS (J&F 48.6) and strong zero-shot transfer to Ref-YouTube-VOS (J&F 47.9) and Ref-DA VIS (J&F 45.4), validating that SAM2's frozen object tokens are language-alignable with light supervision.

Load-bearing premise

The entire pipeline depends on the candidate-generating stage—grid points plus Grounding DINO boxes—actually producing a mask track for every object a sentence might refer to; if the referred object is never prompted, the selection module has nothing to choose, and the method cannot recover.

Editorial extensions

If this is right

  • Because the selector consumes only precomputed tokens, training the full RVOS pipeline takes roughly 7 hours on a single GPU, making such systems feasible without large-scale end-to-end video training.
  • The zero-shot results on Ref-YouTube-VOS and Ref-DA VIS suggest that language-aligned track selection trained on one dataset can transfer to others, positioning it as a reusable building block for language-conditioned video object tasks.
  • The IoU-based pseudo-labeling cleanly separates track quality from selection quality: the selector can be retrained or swapped without regenerating SAM2 tracks, which simplifies future improvements.
  • Since SAM2 is frozen, any future improvement in the underlying segmentation model's object tokens could directly raise SOLA's performance without retraining the selection module.
  • The lightweight design (32.9M trainable parameters) and the use of generic prompts mean the approach can be applied to arbitrary video datasets that provide language annotations, not just the evaluated benchmarks.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The framework's performance upper bound is set by the proposal stage: if neither grid points nor Grounding DINO boxes prompt the referred object, no candidate track exists and selection necessarily fails. A testable extension is to measure candidate recall against ground-truth objects and evaluate SOLA on only the covered subset to isolate selection accuracy.
  • The same frozen-token representation could be reused for other language-conditioned video tasks, such as referring attribute classification or instruction-guided video editing, with the same selection head, since the alignment objective is task-agnostic.
  • The paper's corruption experiments hint that SAM2 tokens are robust to input perturbations, but the mechanism is not isolated; a natural follow-up is to test whether the selection module itself inherits robustness or whether the robustness comes entirely from the frozen backbone.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes SOLA, an RVOS framework that treats SAM2 object tokens as compact video-level object representations. Candidate mask tracks are first generated by prompting a frozen SAM2 with a grid of points and with Grounding DINO bounding boxes sampled every fourth frame; the corresponding SAM2 object pointers are then aligned with RoBERTa text features by a lightweight track selection module. A binary alignment score per track is produced, and tracks above a threshold are merged. Training uses pseudo-labels obtained by thresholding the mIoU between candidate tracks and the ground-truth track, with a BCE loss plus a contrastive alignment loss based on learnable negative anchors. The paper reports state-of-the-art results on MeViS (J&F 48.6 vs 46.4 for DsHmp), zero-shot generalization to Ref-YouTube-VOS and Ref-DA VIS, combined-dataset training, ablations, and analyses of SAM2 object tokens.

Significance. If the central claim holds, the paper makes a useful contribution: it shows that frozen SAM2 object pointers carry enough spatio-temporal object information to support language-guided track selection, and it does so with a very lightweight trainable module (32.9M parameters) that can be trained on a single GPU using precomputed tokens. The design is clearly specified, the ablations are informative, and the zero-shot and combined-dataset results strengthen the evidence that the representation generalizes. The main significance hinges on whether the reported MeViS numbers are a genuine selection-module result rather than an artifact of candidate-track coverage, and on whether the comparison set and evaluation split are complete and clearly identified.

major comments (3)
  1. [3.3, 3.6, Table 1] The proposal-coverage assumption is load-bearing and currently unquantified. Candidate tracks are generated only from grid points and Grounding DINO boxes sampled every fourth frame; any referred object missed by this prompt set has no object token O_i, so Eq. (5) can only assign scores over an incomplete set. The pseudo-labeling in Section 3.6 inherits the same ceiling: if no candidate track reaches mIoU >= tau with the ground-truth track, that expression contributes no positive training example. The paper does not report candidate recall or an oracle upper bound (e.g., selecting the candidate with the highest mIoU). I request an oracle-selection curve and per-expression recall on MeViS so that the reader can determine how much of the 48.6 J&F is attributable to the language-aligned selection module rather than to the proposal stage, and how much is lost on expressions whose referent is absent from the candidates.
  2. [4.1, Table 1] The evaluation split for the main MeViS comparison is not identified. The dataset section states that MeViS has a validation set of 140 videos and a test set of 154 videos, but Table 1 only says “MeViS”. If the reported 48.6 J&F is on the validation set, this must be stated; if it is on the test set, the paper must explain how the test set was obtained and what protocol was used. In either case, the main results should include multiple-seed standard deviations, since the margin over DsHmp is 2.2 J&F and the paper currently reports single-run scores without any indication of variance.
  3. [2, Table 1] The related-work section discusses SAM/SAM2-based RVOS methods (Ref-SAM and AL-RefSAM2) and criticizes their prompt-level use of SAM2, but Table 1 does not compare against them. For the “state-of-the-art on MeViS” claim to be evaluable, the paper should either include the published MeViS results of these methods or explicitly state the reason for their exclusion (for example, if they do not report on MeViS or use additional training data).
minor comments (4)
  1. [Tables 2, 3, A.1] The citation label for ReferFormer is inconsistent: Tables 2, 3, and A.1 cite ReferFormer as [12], but reference [12] is the CLIP paper (Jia et al.), while ReferFormer is reference [28] (Wu et al.).
  2. [Abstract, Section 1] The project page URL differs between the abstract (https://cvlab-kaist.github.io/SOLA) and the introduction (https://github.com/cvlab-kaist/SOLA); please unify.
  3. [4.2] There is a typo in “Nneg = 32for number of negative anchors” (missing space), and the sentence “relies solely exclusively on object tokens” in Section 4.3 is redundant; one of the two adverbs should be removed.
  4. [4.5, Figure 4] The token-similarity analysis would be easier to interpret if the caption stated how the curve was computed: the number of object-token pairs, the binning of mIoU, and whether the pairs are from the same video or across videos.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the track-selection scores are learned network outputs evaluated on held-out data, and the IoU pseudo-labels are a training target, not an inference input.

full rationale

The paper derives its prediction from frozen SAM2 object tokens and RoBERTa text tokens through a learned track selection module (Eqs. 3–5). The alignment scores are genuine network outputs, not a rearrangement of the pseudo-label definition. IoU-based pseudo-labels (Sec. 3.6) are computed from ground-truth mask tracks only during training to supervise binary classification and a contrastive loss; at inference the ground-truth masks are absent, so the predicted selection is not equivalent to the input or to the label by construction. The final J&F scores on MeViS, Ref-YouTube-VOS, and Ref-DA VIS are measured against external annotations and compared with independently trained baselines. No load-bearing self-citation, uniqueness theorem, or ansatz-smuggling chain is present. The candidate-coverage assumption (Sec. 3.3) could limit performance if SAM2/Grounding DINO miss a referred object, but that is an empirical completeness concern, not a circular reduction: the paper does not define its output in terms of its input, and the selection module cannot recover a missing proposal. Hyperparameter choices such as tau=0.5 are standard validation-set tuning rather than fitted inputs renamed as predictions. The derivation is therefore self-contained with respect to circularity.

Assumptions & free parameters 6 free parameters · 5 assumptions · 1 invented entities

The paper's contribution is an alignment module on top of frozen encoders; the load-bearing assumptions are that SAM2 tokens encode sufficient objectness and that the proposal stage covers referable objects. The only learned entity beyond the selection module is the set of negative anchor embeddings for the contrastive loss. Hyperparameters are standard and chosen by hand.

free parameters (6)
  • Selection and pseudo-label threshold tau = 0.5
    Used in Sec 3.5 to accept or reject tracks at inference and in Sec 3.6 to label positive candidates; chosen by hand, no sensitivity analysis reported.
  • Number of negative anchors Nneg = 32
    Set in Sec 4.2 for the alignment loss (Eq 8); chosen by hand.
  • Loss weights lambda1 and lambda2 = 1.0 and 0.3
    Set in Sec 4.2 to combine BCE and alignment losses; chosen by hand.
  • Number of alignment layers L = 2
    Ablation Table A.2b shows 2 layers gives best validation performance; chosen via experimentation.
  • Prompt frame interval = every 4th frame
    Used for prompt mask generation in Sec 4.2; chosen by hand, not swept.
  • NMS IoU threshold for prompt masks = 0.7
    In supplementary Sec D, filters duplicate prompt masks before propagation; chosen by hand.
assumptions (5)
  • domain assumption SAM2 object tokens encode spatiotemporal object information sufficient for language alignment.
    Introduced as a hypothesis in Sec 3.4; supported by token similarity vs mIoU (Fig 4) and PASCAL VOC linear probe (85.3%), but not formally proven.
  • domain assumption Candidate tracks from grid points and Grounding DINO boxes cover every referable object.
    Sec 3.3 and Sec 4.2; if an object is not proposed, the selection module cannot select it, making this assumption load-bearing.
  • domain assumption IoU-based pseudo-labels are a reliable proxy for language-track correspondence.
    Sec 3.6; assumes high mIoU with the ground-truth mask means the track is the referred object, which can fail for similar-looking objects or imprecise masks.
  • domain assumption Lightweight attention and contrastive losses can bridge the modality gap between frozen SAM2 tokens and RoBERTa text features.
    Sec 3.5; the architecture is standard, but its sufficiency for RVOS is an empirical assumption.
  • domain assumption The text encoder RoBERTa provides features adequate for RVOS cues without fine-tuning.
    The paper freezes RoBERTa and notes in the limitations that its training objective differs from RVOS; the assumption is that the features still contain enough motion-relevant information.
invented entities (1)
  • Negative anchor embeddings An
    purpose: Contrastive negative prototypes in the alignment loss (Eq 8), used to push non-matching alignment tokens away.
    These are learnable parameters introduced by the paper; they have no external falsifiable handle and are a standard contrastive-learning component, not a new physical entity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Referring Video Object Segmentation via Language-aligned Track Selection." pith.science (2026). https://pith.science/paper/PXA2AY2R

@misc{pith2026241201136,
  author       = {Pith},
  title        = {Pith review of: Referring Video Object Segmentation via Language-aligned Track Selection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PXA2AY2R}},
  note         = {Machine review of arXiv:2412.01136}
}
read the original abstract

Referring video object segmentation (RVOS) requires tracking and segmenting an object throughout a video according to a given natural language expression, demanding both complex motion understanding and the alignment of visual representations with language descriptions. Given these challenges, the recently proposed Segment Anything Model 2 (SAM2) emerges as a potential candidate due to its ability to generate coherent segmentation mask tracks across video frames, and provide an inherent spatio-temporal objectness in its object token representations. In this paper, we introduce SOLA (Selection by Object Language Alignment), a novel framework that leverages SAM2 object tokens as compact video-level object representations, which are aligned with language features through a lightweight track selection module. To effectively facilitate this alignment, we propose an IoU-based pseudo-labeling strategy, which bridges the modality gap between SAM2 representations with language features. Extensive experiments show that SOLA achieves state-of-the-art performance on the MeViS dataset and demonstrate that SOLA offers an effective solution for RVOS. Our project page is available at: https://cvlab-kaist.github.io/SOLA.

Figures

Figures reproduced from arXiv: 2412.01136 by the authors.

Figure 1
Figure 1. Teaser. Our method effectively bridges the modality gap by aligning the features obtained from fully frozen uni-modal en￾coders: the video segmentation model such as SAM2 [24] and the text encoder such as RoBERTa [20]. By directly leveraging the token representations, our approach achieves lightweight multi￾modal alignment while significantly reducing the number of train￾able parameters. mentation masks across video… view at source ↗
Figure 2
Figure 2. Overall pipeline of the proposed SOLA framework. Our method selects the correct object mask track among candidates via a language-aligned track selection module. We first generate candidate mask tracks and corresponding object tokens from the fully frozen SAM2. These tokens are then aligned with language expressions, producing alignment scores that indicate selection probabilities. Mask tracks with scores above a pr… view at source ↗
Figure 3
Figure 3. Architecture of the language-aligned track selection module, which takes object tokens and text tokens as inputs, align￾ing these representations to effectively capture object dynamics. where ⊗ denotes element-wise multiplication operation, and AvgT (·) represents the mean along the T dimension. Here, Oa, is not only aligned with the language expres￾sion but has also incorporated temporally aggregated mo￾tion inform… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Spatial and motion information in object tokens. The bold line represents the mean similarity, while the shaded region indicates the variance. The results show a certain correlation: as the mIoU between mask tracks increases, the similarity between their associated tok…
Figure 5
Figure 5. Figure 5: Qualitative results of our model on MeViS. SOLA shows its ability to understand complex language expressions. tion between the spatial proximity of masks and the sim￾ilarity of their object tokens. It provides a reasonable in￾dication that object tokens implicitly enco…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. MOVE: Motion-Guided Few-Shot Video Object Segmentation

    cs.CV 2025-07 conditional novelty 7.0 of 10

    MOVE provides a new motion-guided few-shot video object segmentation benchmark, and the proposed DMA baseline outperforms six existing methods across all settings.

  2. Unleashing Hierarchical Reasoning: An LLM-Driven Framework for Training-Free Referring Video Object Segmentation

    cs.CV 2025-09 conditional novelty 6.0 of 10

    A training-free pipeline parses a video query with an LLM, tracks candidates with SAM2, then uses coarse motion plus conditional pose reasoning to segment the referred object.

Reference graph

Works this paper leans on

29 extracted references · 20 canonical work pages · cited by 2 Pith papers

  1. [1]

    Gpt-4 technical report

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ah- mad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774 ,

    arXiv

  2. [2]

    End-to-end referring video object segmentation with mul- timodal transformers

    Adam Botach, Evgenii Zheltonozhskii, and Chaim Baskin. End-to-end referring video object segmentation with mul- timodal transformers. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 4975–4985. IEEE Computer Society, 2022. 2, 6

    work page 2022

  3. [3]

    Vision-language transformer and query generation for refer- ring segmentation

    Henghui Ding, Chang Liu, Suchen Wang, and Xudong Jiang. Vision-language transformer and query generation for refer- ring segmentation. In Proceedings of the IEEE International Conference on Computer Vision, 2021. 6

    work page 2021

  4. [4]

    Mevis: A large-scale benchmark for video segmentation with motion expressions

    Henghui Ding, Chang Liu, Shuting He, Xudong Jiang, and Chen Change Loy. Mevis: A large-scale benchmark for video segmentation with motion expressions. InProceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 2694–2703, 2023. 1, 2, 6, 7

    work page 2023

  5. [5]

    Language-bridged spatial-temporal interaction for referring video object segmentation

    Zihan Ding, Tianrui Hui, Junshi Huang, Xiaoming Wei, Jizhong Han, and Si Liu. Language-bridged spatial-temporal interaction for referring video object segmentation. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4964–4973, 2022. 6

    work page 2022

  6. [6]

    The pascal visual object classes (voc) challenge

    Mark Everingham, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. The pascal visual object classes (voc) challenge. International journal of computer vision, 88:303–338, 2010. 8

    work page 2010

  7. [7]

    Actor and action video segmentation from a sen- tence

    Kirill Gavrilyuk, Amir Ghodrati, Zhenyang Li, and Cees GM Snoek. Actor and action video segmentation from a sen- tence. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 5958–5966. IEEE Computer Society, 2018. 1, 2

    work page 2018

  8. [8]

    Html: Hybrid temporal-scale mul- timodal learning framework for referring video object seg- mentation

    Mingfei Han, Yali Wang, Zhihui Li, Lina Yao, Xiaojun Chang, and Yu Qiao. Html: Hybrid temporal-scale mul- timodal learning framework for referring video object seg- mentation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 13414–13423, 2023. 2

    work page 2023

Show all 29 references
  1. [9]

    Decoupling static and hier- archical motion perception for referring video segmentation

    Shuting He and Henghui Ding. Decoupling static and hier- archical motion perception for referring video segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 13332–13341, 2024. 2, 6, 1, 3, 4, 5

    2024

  2. [10]

    Benchmarking neu- ral network robustness to common corruptions and perturba- tions

    Dan Hendrycks and Thomas Dietterich. Benchmarking neu- ral network robustness to common corruptions and perturba- tions. arXiv preprint arXiv:1903.12261, 2019. 1

    1903 arXiv

  3. [11]

    Unleashing the temporal-spatial reasoning capacity of gpt for training-free audio and language referenced video object segmentation

    Shaofei Huang, Rui Ling, Hongyu Li, Tianrui Hui, Zongheng Tang, Xiaoming Wei, Jizhong Han, and Si Liu. Unleashing the temporal-spatial reasoning capacity of gpt for training-free audio and language referenced video object segmentation. arXiv preprint arXiv:2408.15876, 2024. 2

    2024 arXiv

  4. [12]

    Scaling up visual and vision-language representa- tion learning with noisy text supervision

    Chao Jia, Yinfei Yang, Ye Xia, Yi-Ting Chen, Zarana Parekh, Hieu Pham, Quoc Le, Yun-Hsuan Sung, Zhen Li, and Tom Duerig. Scaling up visual and vision-language representa- tion learning with noisy text supervision. In International conference on machine learning , pages 4904–49...

  5. [13]

    Video object segmentation with language referring expressions

    Anna Khoreva, Anna Rohrbach, and Bernt Schiele. Video object segmentation with language referring expressions. In ACCV, 2018. 2, 6

    2018

  6. [14]

    Video object segmentation with language referring expressions

    Anna Khoreva, Anna Rohrbach, and Bernt Schiele. Video object segmentation with language referring expressions. In 14th Asian Conference on Computer Vision , pages 123–141. Springer, 2019. 1

    2019

  7. [15]

    Segment any- thing

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C Berg, Wan-Yen Lo, et al. Segment any- thing. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 4015–4026, 2023. 1, 2

    2023

  8. [16]

    Referring image seg- mentation via recurrent refinement networks

    Ruiyu Li, Kaican Li, Yi-Chun Kuo, Michelle Shu, Xiaojuan Qi, Xiaoyong Shen, and Jiaya Jia. Referring image seg- mentation via recurrent refinement networks. In Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 5745–5753, 2018. 2

    2018

  9. [17]

    Robust referring video object segmentation with cyclic structural consensus

    Xiang Li, Jinglu Wang, Xiaohao Xu, Xiao Li, Bhiksha Raj, and Yan Lu. Robust referring video object segmentation with cyclic structural consensus. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 22236– 22245, 2023. 2

    2023

  10. [18]

    Refsam: Efficiently adapting segmenting any- thing model for referring video object segmentation

    Yonglin Li, Jing Zhang, Xiao Teng, Long Lan, and Xin- wang Liu. Refsam: Efficiently adapting segmenting any- thing model for referring video object segmentation. arXiv preprint arXiv:2307.00997, 2023. 2

    2023 arXiv

  11. [19]

    Grounding dino: Marrying dino with grounded pre-training for open-set object detection

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Chunyuan Li, Jianwei Yang, Hang Su, Jun Zhu, et al. Grounding dino: Marrying dino with grounded pre-training for open-set object detection. arXiv preprint arXiv:2303.05499, 2023. 2, 6, 7

    2023 arXiv

  12. [20]

    Roberta: A robustly optimized BERT pretraining approach

    Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettle- moyer, and Veselin Stoyanov. Roberta: A robustly optimized BERT pretraining approach. CoRR, abs/1907.11692, 2019. 1, 6, 2

    1907 arXiv

  13. [21]

    Temporally consistent referring video object segmentation with hybrid memory

    Bo Miao, Mohammed Bennamoun, Yongsheng Gao, Mubarak Shah, and Ajmal Mian. Temporally consistent referring video object segmentation with hybrid memory. IEEE Transactions on Circuits and Systems for Video Tech- nology, 2024. 2, 6

    2024

  14. [22]

    Efficient non- maximum suppression

    Alexander Neubeck and Luc Van Gool. Efficient non- maximum suppression. In 18th international conference on pattern recognition (ICPR’06), pages 850–855. IEEE, 2006. 6

    2006

  15. [23]

    A benchmark dataset and evaluation methodology for video object segmentation

    F Perazzi, J Pont-Tuset, B McWilliams, L Van Gool, M Gross, and A Sorkine-Hornung. A benchmark dataset and evaluation methodology for video object segmentation. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 724–732. IEEE, 2016. 6

    2016

  16. [24]

    Sam 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman R¨adle, Chloe Rolland, Laura Gustafson, et al. Sam 2: Segment anything in images and videos. arXiv preprint arXiv:2408.00714, 2024. 1, 2, 3, 4, 6 9

    2024 arXiv

  17. [25]

    Urvos: Unified referring video object segmentation network with a large-scale benchmark

    Seonguk Seo, Joon-Young Lee, and Bohyung Han. Urvos: Unified referring video object segmentation network with a large-scale benchmark. In European Conference on Com- puter Vision, pages 208–223, 2020. 1, 6, 7

    2020

  18. [26]

    Attention is all you need

    A Vaswani. Attention is all you need. Advances in Neural Information Processing Systems, 2017. 4

    2017

  19. [27]

    Data-efficient mul- timodal fusion on a single gpu

    No ¨el V ouitsis, Zhaoyan Liu, Satya Krishna Gorti, Valentin Villecroze, Jesse C Cresswell, Guangwei Yu, Gabriel Loaiza-Ganem, and Maksims V olkovs. Data-efficient mul- timodal fusion on a single gpu. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Rec...

    2024

  20. [28]

    Language as queries for referring video object segmen- tation

    Jiannan Wu, Yi Jiang, Peize Sun, Zehuan Yuan, and Ping Luo. Language as queries for referring video object segmen- tation. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 4964–4974. IEEE,

    2022

  21. [2022]

    Going right

    2, 6 10 Referring Video Object Segmentation via Language-aligned Track Selection Supplementary Material A. Additional qualitative results Qualitative results on MeViS. Figure A.1 presents the qualitative results on MeViS [4], comparing the perfor- mance of DsHmp [9] with SOLA....

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.