Pith. sign in

REVIEW 4 major objections 6 minor 90 references

CLDTracker: A Comprehensive Language Description for Visual Tracking

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read CLDTracker claims that a comprehensive bag of textual descriptions, updated over time, lets vision-language trackers outdo 38 prior methods on six benchmarks.

desk verdict Solid contribution with strong ablations, but the GOT-10k protocol contradiction and missing code/data keep me from trusting the headline SOTA claim. read the letter →

arxiv 2505.23704 v1 pith:TFDRW42V submitted 2025-05-29 cs.CV cs.AI

Mohamad Alansari , Sajid Javed , Iyyakutti Iyappan Ganapathi , Sara Alansari , Muzammal Naseer This is my paper
classification cs.CVcs.AI
keywords visualobjecttrackingvision-languagemodelscomprehensivelanguagedescriptionbagoftextualdescriptionstemporaltextfeatureupdatepromptadapterGPT-4VCLIP
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

CLDTracker claims that visual object tracking improves when the target is given a comprehensive, multi-perspective text description instead of the single phrase most benchmarks provide. From the first frame, the method builds a bag of textual descriptions covering class name, attribute phrase, GPT-4V-generated caption, and semantic/context enrichment, then aligns these texts with visual features and adaptively updates their weights over time. Across six benchmarks, the paper reports the best success score on LaSOT (74.0 AUC), the best normalized precision on all five language-annotated benchmarks, and the best AO (77.5) and SR0.75 (75.6) on GOT-10k among the 38 compared methods. The central claim is that a static, single-sentence description is a bottleneck in vision-language tracking, and that comprehensiveness plus temporal updating removes that bottleneck.

What carries the argument

The carrying object is the bag of textual descriptions $\mathcal{B}_t$, built offline from the first frame: the best CLIP-matched class and attribute phrase, a GPT-4V caption, and a semantic/context enrichment containing synonyms, perturbed forms, task-specific phrases, and a high-level concept. Two learned modules carry the argument: a Prompt Adapter (CoCoOp-style conditioned context vectors) selects the most relevant text per frame, and the Temporal Text Feature Update Mechanism (TTFUM) computes attention weights from the absolute difference between the exemplar text and the average of past search-region texts, producing $T_{\mathrm{att}}$. The final correlation $\mathrm{Corr}(F_{\mathrm{frame}}, \tilde{T}_e) = (1 + T_{\mathrm{att}}(\tilde{T}_e)) \ast F_{\mathrm{frame}}$ uses the text feature as a convolution kernel over the visual feature map, which is what converts language into a spatially localized target prediction.

What would settle it

Take a set of sequences where a target changes appearance irreversibly after initialization (for example, a player changes jersey, or a vehicle is repainted) and a same-category distractor appears later. If CLDTracker's success on that subset is no better than CiteTracker's, or if injecting a random first-frame description into the bag produces a collapse comparable to the L0 zero-text ablation, then the claim that comprehensiveness plus TTFUM repairs drift is not supported.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that many-to-one text–image alignment beats one-to-one alignment in tracking. CLDTracker generates a bag of texts from the first frame, selects the most relevant one per frame with a learned prompt adapter, and continuously reweights past textual features through a temporal update mechanism; the chosen text embedding is then used as a convolution kernel against the visual feature map. The paper reports consistent gains over CiteTracker, its vision-language baseline, and argues that these gains show richer language inputs supply disambiguating cues where short benchmark phrases fail. It also reports strong generalization on VastTrack, including on classes absent from the constructed dictionaries, positioning the method as capable of tracking novel categories without retraining.

Load-bearing premise

The load-bearing premise is that a bag of texts generated once from the first frame, together with TTFUM's averaging of past frame features, remains a faithful description of the target for the whole sequence; if the first-frame descriptions are wrong or go stale during long occlusion, the correlation kernel in Eq. (9) is guided by misleading language.

Editorial extensions

If this is right

  • If correct, language-guided tracking no longer needs hand-written per-frame sentences: one offline-generated bag plus a temporal update matches or beats manual annotations on TNL2K and OTB99-Lang.
  • The normalized-precision gains imply the method is more robust to annotation style, not just to target appearance, so it could absorb noisy or paraphrased descriptions in deployed systems.
  • The computational path of pre-encoded text features reused across frames implies comprehensive text can be added at 35.31 FPS with 54.25G MACs, roughly a third of the cost of the CiteTracker baseline.
  • Generalization to 95 unseen VastTrack classes suggests the pipeline can track novel categories without retraining, as long as the prompt adapter can condition on visual features alone.

Reading between the lines

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

  • A testable extension follows from the paper's own failure case: if the bag of descriptions were regenerated periodically instead of only from the first frame, long-occlusion sequences should improve, which is exactly what the limitation section predicts.
  • The TTFUM attention formula treats textual grounding as a dynamical state, suggesting the same averaging-plus-difference scheme could be applied to streaming open-vocabulary detection or re-identification where object appearance drifts.
  • The dictionary coverage constraint of 940 classes and 23,899 attributes could be relaxed by replacing fixed vocabularies with on-the-fly retrieval, which might extend the reported gains to truly arbitrary objects.
  • If the paper's mechanism is the real driver, then holding the visual branch fixed and varying only the richness of text inputs should produce monotone accuracy gains, which is an ablation not reported in the current paper.
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

4 major / 6 minor

Summary. CLDTracker proposes a vision-language tracker that extends the OSTrack/CiteTracker architecture with a bag of textual descriptions (Bt) built offline from the first frame via predefined class/attribute dictionaries, CLIP matching, GPT-4V captions, and semantic/contextual enrichment, plus a Temporal Text Feature Update Mechanism (TTFUM) that averages search-region text features over a temporal window. The paper reports results on LaSOT, LaSOText, TrackingNet, TNL2K, OTB99-Lang, GOT-10k, and VastTrack, claiming state-of-the-art or top-tier performance, and provides extensive ablations of each component, robustness tests to noisy text, and an efficiency analysis.

Significance. The idea of using a diverse, multi-perspective text bag rather than a single static phrase is timely and the experimental infrastructure is substantial: 38 competitors, multiple benchmarks, component-wise ablations, robustness checks, and a useful unseen-class generalization test on VastTrack (Table 15). The efficiency improvement over CiteTracker (35.31 vs. 15.35 FPS with lower MACs) is also a concrete practical strength. However, the 'SOTA' claim is not uniform: the method is second on several benchmarks and metrics. More importantly, the GOT-10k training-protocol ambiguity, the absence of error bars, and the fact that code and Bt are not yet available prevent the results from being fully verified at this stage. If the protocol issue is resolved and the artifacts are released, the work would be a solid contribution to VLM-based tracking.

major comments (4)
  1. [Sec. 4.1 and Sec. 4.2, Table 1] Section 4.1 states that CLDTracker was trained on GOT-10k, TrackingNet, LaSOT, and COCO2017, while Section 4.2 acknowledges that GOT-10k's official protocol prohibits external training data and only says the model was evaluated 'under similar conditions.' If the reported GOT-10k AO=77.5 and SR0.75=75.6 were produced by the mixed-data model, they are not comparable to protocol-compliant trackers such as CiteTracker (AO 74.7) or HIPTrack (AO 77.4), and the claim of 'highest scores in both AO and SR0.75' is unsupported. The authors must disclose the exact training data used for the GOT-10k model and, if external data were used, retrain under the official protocol or withdraw the GOT-10k claims.
  2. [Table 1] The margins separating CLDTracker from the runner-up are small on several decisive metrics (e.g., GOT-10k AO 77.5 vs. HIPTrack 77.4; LaSOText S 53.1 vs. VLT_OST-384 53.5; OTB99-Lang S 77.8 vs. VLT_OST-384 78.2). Single-run evaluations with no variance estimates or significance tests cannot support 'consistently top-tier' or 'best' claims. Please report mean and standard deviation over at least three seeds, or provide official leaderboard significance analysis if available.
  3. [Sec. 3.2.1 and Sec. 4.9] The abstract and introduction claim 'temporally-adaptive vision-language representations,' but the Bt is constructed once from the first frame and TTFUM only averages features from the same static bag over a temporal window (Eq. 8). Section 4.9 explicitly acknowledges that 'the use of a static Bt, generated only from the first frame, limits adaptability to significant appearance changes over time.' This limitation weakens the temporal-modeling contribution as stated; either the method should regenerate or augment descriptions online, or the claims should be explicitly scoped to 'temporally refined text features from a static bag.'
  4. [Sec. 3.2.1(e) and Eq. (6)] The Bt is filtered using CLIP similarity to the first-frame visual feature FI with threshold tau=0.8, and the same CLIP text encoder and FI-derived features are used during inference to select prompts via Eq. (6). This creates a potential selection bias: the Bt is optimized to match CLIP's own embedding rather than to be semantically faithful. A control experiment using human-written captions or an independent captioning model not filtered by CLIP would help show that the gains come from comprehensive language grounding rather than from exploiting CLIP's self-consistency.
minor comments (6)
  1. [Sec. 2.1] There is a typo in the introduction: 'Thereofore' should be 'Therefore.'
  2. [Sec. 3.2.1(d) and Sec. 3.2.1(e)] The symbol tau is used both as the synonym-retrieval threshold in Eq. (2) and as the filtering threshold in Sec. 3.2.1(e); please rename one to avoid confusion.
  3. [Algorithm 1] Lines 6-7 write the class and attribute selection as a joint argmax over (j,k), but the objective adds two separate similarity terms; clarify that the class and attribute are selected independently.
  4. [Sec. 4.2] The abstract says six benchmarks, but Section 4.2 also lists VastTrack as an evaluation benchmark; please reconcile the count or describe VastTrack as an additional generalization test.
  5. [Abstract and Sec. 4.9] The abstract and the Research Data Availability statement say code and models are publicly available, while Section 4.9 says they 'will be made available'; please make the GitHub repository and Bt accessible at the time of publication or state a concrete release date.
  6. [Table 4] The column header 'Baseline VL Model (WS)' is visually ambiguous because the same header spans all WS blocks; please add explicit WS labels under each block or restructure the table so the w/ and w/o Prompt Adapter configurations are immediately clear.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the claimed SOTA results are empirical benchmark measurements, and the CLIP-based text generation/selection loop does not force the tracking output by construction.

full rationale

The central claim is an empirical performance result, not a derived quantity. The textual bag Bt is built from the first frame via CLIP matching (Eq. 1), GPT-4V captioning, and a CLIP-based validation filter (Sec. 3.2.1(e): "Descriptions that meet the semantic alignment constraint sim(FI, di) ≥ τ, with τ = 0.8, are retained"). The same CLIP encoder also scores descriptions at inference (Eq. 6). This is self-referential in a weak sense: CLIP is used on both sides of text selection, and the exemplar image is the same first frame used to generate Bt. However, this does not make the tracker's output a tautology. The correlation head in Eq. 9 convolves the selected text kernel with search-region visual features and must still localize the target in novel frames; the method demonstrably can fail, as the paper's own failure case and the OV/full-occlusion results show. The SOTA numbers on LaSOT, TrackingNet, GOT-10k, etc. are benchmark measurements taken from an evaluation pipeline, not quantities defined by the method's equations. No load-bearing self-citation or imported uniqueness theorem is present; the only self-citation (Alansari et al. 2024) is an application-level reference and does not support the method's core architecture. The GOT-10k training-protocol concern and the tuning of λiou, λL1, and w on LaSOT are correctness/benchmark-integrity risks rather than circular reductions, because the reported AUC is not equal to the fitted hyperparameters by construction.

Assumptions & free parameters 8 free parameters · 4 assumptions · 0 invented entities

The central claim is empirical, not a formal derivation. The main load-bearing inputs are the pretrained CLIP and GPT-4V models, the hand-merged dictionaries, and the hyperparameters tuned on the benchmarks. There are no new physical entities.

free parameters (8)
  • Description filtering threshold tau = 0.8
    Text descriptions are kept only if CLIP similarity to the first-frame image is at least 0.8 (Section 3.2.1e). This value is chosen without a reported sensitivity analysis.
  • Number of synonyms n = 10
    R(w) returns the first 10 synonyms in Eq. 3, a design choice.
  • Semantic perturbation probability alpha = Not specified
    Eq. 4 defines a perturbation probability alpha in [0,1], but the paper does not state the value used.
  • Temporal window size WS = 5
    TTFUM averages over the past n=5 frames based on the ablation in Table 4.
  • Loss weights lambda_iou, lambda_L1 = 2 and 5
    Set empirically in Section 3.6; ablation in Table 13 confirms they are optimal on LaSOT.
  • Hanning window weight w = 0.49
    Set empirically in Section 3.7; ablation in Table 14.
  • Number of class dictionary entries = 940
    Size of the merged class dictionary.
  • Number of attribute dictionary entries = 23,899
    Size of the merged attribute dictionary.
assumptions (4)
  • domain assumption CLIP embeddings provide enough semantic alignment to select correct class/attribute and validate descriptions.
    Used throughout Section 3.2.1; the whole text branch depends on CLIP matching accuracy.
  • domain assumption GPT-4V generates accurate and useful descriptions when prompted with the first-frame bounding box.
    GPT-4V descriptions are added to Bt (Section 3.2.1c); the paper shows GPT-4V can be misled (Fig. 8j), so this is load-bearing.
  • domain assumption The merged dictionaries and WordNet synonyms cover target objects in the benchmarks.
    Dictionaries are built from six datasets; for unseen classes the assumption is weaker.
  • standard math The benchmarks' evaluation protocols and metrics are applied correctly.
    Standard tracking evaluation; no verification of the server runs.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CLDTracker: A Comprehensive Language Description for Visual Tracking." pith.science (2026). https://pith.science/paper/TFDRW42V

@misc{pith2026250523704,
  author       = {Pith},
  title        = {Pith review of: CLDTracker: A Comprehensive Language Description for Visual Tracking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TFDRW42V}},
  note         = {Machine review of arXiv:2505.23704}
}
read the original abstract

VOT remains a fundamental yet challenging task in computer vision due to dynamic appearance changes, occlusions, and background clutter. Traditional trackers, relying primarily on visual cues, often struggle in such complex scenarios. Recent advancements in VLMs have shown promise in semantic understanding for tasks like open-vocabulary detection and image captioning, suggesting their potential for VOT. However, the direct application of VLMs to VOT is hindered by critical limitations: the absence of a rich and comprehensive textual representation that semantically captures the target object's nuances, limiting the effective use of language information; inefficient fusion mechanisms that fail to optimally integrate visual and textual features, preventing a holistic understanding of the target; and a lack of temporal modeling of the target's evolving appearance in the language domain, leading to a disconnect between the initial description and the object's subsequent visual changes. To bridge these gaps and unlock the full potential of VLMs for VOT, we propose CLDTracker, a novel Comprehensive Language Description framework for robust visual Tracking. Our tracker introduces a dual-branch architecture consisting of a textual and a visual branch. In the textual branch, we construct a rich bag of textual descriptions derived by harnessing the powerful VLMs such as CLIP and GPT-4V, enriched with semantic and contextual cues to address the lack of rich textual representation. Experiments on six standard VOT benchmarks demonstrate that CLDTracker achieves SOTA performance, validating the effectiveness of leveraging robust and temporally-adaptive vision-language representations for tracking. Code and models are publicly available at: https://github.com/HamadYA/CLDTracker

Figures

Figures reproduced from arXiv: 2505.23704 by the authors.

Figure 1
Figure 1. VOT performance comparison in terms of Area Under the Curve (AUC) score between the CiteTracker Li et al. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Comparison of the visual-textual alignment process in the VLM-based trackers. (a) show the traditional visual-textual [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Schematic illustration of our proposed Comprehensive Language Description framework for Visual Tracking (CLD [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: System diagram illustrating the process of building a [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: AUC scores of different attributes on (a) LaSOT, and (b) LaSOT [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]
Figure 6
Figure 6. Figure 6: Results visualization of different trackers. The comparison shows that our CLDTracker could perform robust tracking [PITH_FULL_IMAGE:figures/full_fig_p022_6.png]
Figure 7
Figure 7. Figure 7: Visual results of various trackers under different input language descriptions. The comparison highlights the robustness [PITH_FULL_IMAGE:figures/full_fig_p023_7.png]
Figure 8
Figure 8. Figure 8: Activation visualizations of CLDTracker and CiteTracker using GradCAM Selvaraju et al. (2017) on two examples: [PITH_FULL_IMAGE:figures/full_fig_p024_8.png]
Figure 9
Figure 9. Figure 9: Visual results demonstrating the robustness of our proposed TTFUM against occlusion. The sequence progresses from [PITH_FULL_IMAGE:figures/full_fig_p028_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

90 extracted references · 57 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  2. [2]

    , author Abdul Hay, O

    author Alansari, M. , author Abdul Hay, O. , author Alansari, S. , author Javed, S. , author Shoufan, A. , author Zweiri, Y. , & author Werghi, N. ( year 2024 ). title Drone-person tracking in uniform appearance crowd: A new dataset . journal Scientific Data \/ , volume 11 \/ , pages 15

    2024

  3. [3]

    , author Naseer, M

    author Awais, M. , author Naseer, M. , author Khan, S. , author Anwer, R. M. , author Cholakkal, H. , author Shah, M. , author Yang, M.-H. , & author Khan, F. S. ( year 2025 ). title Foundation models defining a new era in vision: a survey and outlook . journal IEEE Transactions on Pattern Analysis and Machine Intelligence \/ ,

    2025

  4. [4]

    , author Valmadre, J

    author Bertinetto, L. , author Valmadre, J. , author Henriques, J. F. , author Vedaldi, A. , & author Torr, P. H. ( year 2016 ). title Fully-convolutional siamese networks for object tracking . In booktitle Computer Vision--ECCV 2016 Workshops: Amsterdam, The Netherlands, October 8-10 and 15-16, 2016, Proceedings, Part II 14 \/ (pp. pages 850--865 ). orga...

    2016

  5. [5]

    , author Danelljan, M

    author Bhat, G. , author Danelljan, M. , author Gool, L. V. , & author Timofte, R. ( year 2019 ). title Learning discriminative model prediction for tracking . In booktitle Proceedings of the IEEE/CVF international conference on computer vision \/ (pp. pages 6182--6191 )

    2019

  6. [6]

    O'Reilly Media, Inc

    author Bird, S. , author Klein, E. , & author Loper, E. ( year 2009 ). title Natural language processing with Python: analyzing text with the natural language toolkit \/ . publisher " O'Reilly Media, Inc."

    2009

  7. [7]

    author Bolme, D. S. , author Beveridge, J. R. , author Draper, B. A. , & author Lui, Y. M. ( year 2010 ). title Visual object tracking using adaptive correlation filters . In booktitle 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition \/ (pp. pages 2544--2550 ). :10.1109/CVPR.2010.5539960

    arXiv 2010

  8. [8]

    , author Liu, Q

    author Cai, W. , author Liu, Q. , & author Wang, Y. ( year 2024 ). title Hiptrack: Visual tracking with historical prompts . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition \/ (pp. pages 19258--19267 )

    2024

Show all 90 references
  1. [9]

    , author Wang, X

    author Chen, F. , author Wang, X. , author Zhao, Y. , author Lv, S. , & author Niu, X. ( year 2022 ). title Visual object tracking: A survey . journal Computer Vision and Image Understanding \/ , volume 222 \/ , pages 103508

    2022

  2. [10]

    , author Peng, H

    author Chen, X. , author Peng, H. , author Wang, D. , author Lu, H. , & author Hu, H. ( year 2023 ). title Seqtrack: Sequence to sequence learning for visual object tracking . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition \/ (pp...

    2023

  3. [11]

    , author Yan, B

    author Chen, X. , author Yan, B. , author Zhu, J. , author Wang, D. , author Yang, X. , & author Lu, H. ( year 2021 ). title Transformer tracking . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition \/ (pp. pages 8126--8135 )

    2021

  4. [12]

    , author Zhong, B

    author Chen, Z. , author Zhong, B. , author Li, G. , author Zhang, S. , & author Ji, R. ( year 2020 ). title Siamese box adaptive network for visual tracking . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) \/

    2020

  5. [13]

    , author Song, L

    author Cheng, T. , author Song, L. , author Ge, Y. , author Liu, W. , author Wang, X. , & author Shan, Y. ( year 2024 ). title Yolo-world: Real-time open-vocabulary object detection . In booktitle IEEE CVPR \/

    2024

  6. [14]

    , author Kim, T

    author Cho, H. , author Kim, T. , author Jeong, Y. , & author Yoon, K.-J. ( year 2024 ). title A benchmark dataset for event-guided human pose estimation and tracking in extreme conditions . In editor A. Globerson , editor L. Mackey , editor D. Belgrave , editor A. Fan , edito...

    2024

  7. [15]

    , author Jiang, C

    author Cui, Y. , author Jiang, C. , author Wang, L. , & author Wu, G. ( year 2022 ). title Mixformer: End-to-end tracking with iterative mixed attention . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition \/ (pp. pages 13608--13618 )

    2022

  8. [16]

    , author Jiang, C

    author Cui, Y. , author Jiang, C. , author Wu, G. , & author Wang, L. ( year 2024 ). title Mixformer: End-to-end tracking with iterative mixed attention . journal IEEE Transactions on Pattern Analysis and Machine Intelligence \/ , volume 46 \/ , pages 4129--4146 . :10.1109/TPA...

    2024

  9. [17]

    , author Bhat, G

    author Danelljan, M. , author Bhat, G. , author Khan, F. S. , & author Felsberg, M. ( year 2019 ). title Atom: Accurate tracking by overlap maximization . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition \/ (pp. pages 4660--4669 )

    2019

  10. [18]

    , author Bhat, G

    author Danelljan, M. , author Bhat, G. , author Shahbaz Khan, F. , & author Felsberg, M. ( year 2017 ). title Eco: Efficient convolution operators for tracking . In booktitle Proceedings of the IEEE conference on computer vision and pattern recognition \/ (pp. pages 6638--6646 )

    2017

  11. [19]

    , author Gool, L

    author Danelljan, M. , author Gool, L. V. , & author Timofte, R. ( year 2020 ). title Probabilistic regression for visual tracking . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition \/ (pp. pages 7183--7192 )

    2020

  12. [20]

    , author Chang, M.-W

    author Devlin, J. , author Chang, M.-W. , author Lee, K. , & author Toutanova, K. ( year 2019 ). title Bert: Pre-training of deep bidirectional transformers for language understanding . In booktitle Proceedings of the 2019 conference of the North American chapter of the associ...

    2019

  13. [21]

    , author Beyer, L

    author Dosovitskiy, A. , author Beyer, L. , author Kolesnikov, A. , author Weissenborn, D. , author Zhai, X. , author Unterthiner, T. , author Dehghani, M. , author Minderer, M. , author Heigold, G. , author Gelly, S. et al. ( year 2020 ). title An image is worth 16x16 words: ...

    2020 arXiv

  14. [22]

    , author Bai, H

    author Fan, H. , author Bai, H. , author Lin, L. , author Yang, F. , author Chu, P. , author Deng, G. , author Yu, S. , author Harshit , author Huang, M. , author Liu, J. et al. ( year 2021 ). title Lasot: A high-quality large-scale single object tracking benchmark . journal I...

    2021

  15. [23]

    , author Lin, L

    author Fan, H. , author Lin, L. , author Yang, F. , author Chu, P. , author Deng, G. , author Yu, S. , author Bai, H. , author Xu, Y. , author Liao, C. , & author Ling, H. ( year 2019 ). title Lasot: A high-quality benchmark for large-scale single object tracking . In booktitl...

    2019

  16. [24]

    , author Ablavsky, V

    author Feng, Q. , author Ablavsky, V. , author Bai, Q. , author Li, G. , & author Sclaroff, S. ( year 2020 ). title Real-time visual object tracking with natural language description . In booktitle 2020 IEEE Winter Conference on Applications of Computer Vision (WACV) \/ (pp. p...

    2020

  17. [25]

    , author Ablavsky, V

    author Feng, Q. , author Ablavsky, V. , author Bai, Q. , & author Sclaroff, S. ( year 2021 ). title Siamese natural language tracker: Tracking by natural language descriptions with siamese trackers . In booktitle 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognit...

    2021

  18. [26]

    , author Zhou, C

    author Gao, S. , author Zhou, C. , author Ma, C. , author Wang, X. , & author Yuan, J. ( year 2022 ). title Aiatrack: Attention in attention for transformer visual tracking . In booktitle European Conference on Computer Vision \/ (pp. pages 146--164 ). organization Springer

    2022

  19. [27]

    , author Zhou, C

    author Gao, S. , author Zhou, C. , & author Zhang, J. ( year 2023 ). title Generalized relation modeling for transformer tracking . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition \/ (pp. pages 18686--18695 )

    2023

  20. [28]

    , author Dubey, A

    author Grattafiori, A. , author Dubey, A. , author Jauhri, A. , author Pandey, A. , author Kadian, A. , author Al-Dahle, A. , author Letman, A. , author Mathur, A. , author Schelten, A. , author Vaughan, A. et al. ( year 2024 ). title The llama 3 herd of models . journal arXiv...

    2024 arXiv

  21. [29]

    , author Zhang, Z

    author Guo, M. , author Zhang, Z. , author Fan, H. , & author Jing, L. ( year 2022 ). title Divert more attention to vision-language tracking . In editor S. Koyejo , editor S. Mohamed , editor A. Agarwal , editor D. Belgrave , editor K. Cho , & editor A. Oh (Eds.), booktitle A...

    2022

  22. [30]

    , author Zhang, Z

    author Guo, M. , author Zhang, Z. , author Jing, L. , author Ling, H. , & author Fan, H. ( year 2024 ). title Divert more attention to vision-language object tracking . journal IEEE Transactions on Pattern Analysis and Machine Intelligence \/ , (pp. pages 1--18 )

    2024

  23. [31]

    , author Chen, X

    author He, K. , author Chen, X. , author Xie, S. , author Li, Y. , author Doll \'a r, P. , & author Girshick, R. ( year 2022 ). title Masked autoencoders are scalable vision learners . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recogniti...

    2022

  24. [32]

    author Henriques, J. F. , author Caseiro, R. , author Martins, P. , & author Batista, J. ( year 2014 ). title High-speed tracking with kernelized correlation filters . journal IEEE transactions on pattern analysis and machine intelligence \/ , volume 37 \/ , pages 583--596

    2014

  25. [33]

    , author Zhao, X

    author Huang, L. , author Zhao, X. , & author Huang, K. ( year 2021 ). title Got-10k: A large high-diversity benchmark for generic object tracking in the wild . journal IEEE Transactions on Pattern Analysis and Machine Intelligence \/ , volume 43 \/ , pages 1562--1577 . :10.11...

    2021

  26. [34]

    , author Danelljan, M

    author Javed, S. , author Danelljan, M. , author Khan, F. S. , author Khan, M. H. , author Felsberg, M. , & author Matas, J. ( year 2022 ). title Visual object tracking with discriminative filters and siamese networks: a survey and outlook . journal IEEE TPAMI \/ , volume 45 \...

    2022

  27. [35]

    author Khattak, M. U. , author Rasheed, H. , author Maaz, M. , author Khan, S. , & author Khan, F. S. ( year 2023 ). title Maple: Multi-modal prompt learning . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition \/ (pp. pages 19113--19122 )

    2023

  28. [36]

    , author Mintun, E

    author Kirillov, A. , author Mintun, E. , author Ravi, N. , author Mao, H. , author Rolland, C. , author Gustafson, L. , author Xiao, T. , author Whitehead, S. , author Berg, A. C. , author Lo, W.-Y. et al. ( year 2023 ). title Segment anything . In booktitle Proceedings of th...

    2023

  29. [37]

    , author Kokul, T

    author Kugarajeevan, J. , author Kokul, T. , author Ramanan, A. , & author Fernando, S. ( year 2023 ). title Transformers in single object tracking: An experimental survey . journal IEEE Access \/ , volume 11 \/ , pages 80297--80326 . :10.1109/ACCESS.2023.3298440

    2023

  30. [38]

    , & author Deng, J

    author Law, H. , & author Deng, J. ( year 2018 ). title Cornernet: Detecting objects as paired keypoints . In booktitle Proceedings of the European conference on computer vision (ECCV) \/ (pp. pages 734--750 )

    2018

  31. [39]

    , author Wu, W

    author Li, B. , author Wu, W. , author Wang, Q. , author Zhang, F. , author Xing, J. , & author Yan, J. ( year 2019 ). title Siamrpn++: Evolution of siamese visual tracking with very deep networks . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pat...

    2019

  32. [40]

    , author Yan, J

    author Li, B. , author Yan, J. , author Wu, W. , author Zhu, Z. , & author Hu, X. ( year 2018 ). title High performance visual tracking with siamese region proposal network . In booktitle Proceedings of the IEEE conference on computer vision and pattern recognition \/ (pp. pag...

    2018

  33. [41]

    , author Hong, D

    author Li, C. , author Hong, D. , author Zhang, B. , author Li, Y. , author Camps-Valls, G. , author Zhu, X. X. , & author Chanussot, J. ( year 2025 a ). title Urbansam: Learning invariance-inspired adapters for segment anything models in urban construction . journal arXiv pre...

    2025 arXiv

  34. [42]

    , author Li, D

    author Li, J. , author Li, D. , author Savarese, S. , & author Hoi, S. ( year 2023 a ). title Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models . In booktitle International conference on machine learning \/ (pp. pages 19730-...

    2023

  35. [43]

    , author Selvaraju, R

    author Li, J. , author Selvaraju, R. , author Gotmare, A. , author Joty, S. , author Xiong, C. , & author Hoi, S. C. H. ( year 2021 ). title Align before fuse: Vision and language representation learning with momentum distillation . journal Advances in neural information proce...

    2021

  36. [44]

    , author Huang, Y

    author Li, X. , author Huang, Y. , author He, Z. , author Wang, Y. , author Lu, H. , & author Yang, M.-H. ( year 2023 b ). title Citetracker: Correlating image and text for visual tracking . In booktitle Proceedings of the IEEE/CVF International Conference on Computer Vision \...

    2023

  37. [45]

    , author Li, C

    author Li, X. , author Li, C. , author Ghamisi, P. , & author Hong, D. ( year 2025 b ). title Fleximo: A flexible remote sensing foundation model . journal arXiv preprint arXiv:2503.23844 \/ ,

    2025 arXiv

  38. [46]

    , author Yu, J

    author Li, Y. , author Yu, J. , author Cai, Z. , & author Pan, Y. ( year 2022 ). title Cross-modal target retrieval for tracking by natural language . In booktitle 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) \/ (pp. pages 4927--4936 )....

    2022

  39. [47]

    , author Tao, R

    author Li, Z. , author Tao, R. , author Gavves, E. , author Snoek, C. G. M. , & author Smeulders, A. W. M. ( year 2017 ). title Tracking by natural language specification . In booktitle 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) \/ (pp. pages 7350--...

    2017 doi

  40. [48]

    , author Fan, H

    author Lin, L. , author Fan, H. , author Zhang, Z. , author Xu, Y. , & author Ling, H. ( year 2022 ). title Swintrack: A simple and strong baseline for transformer tracking . journal Advances in Neural Information Processing Systems \/ , volume 35 \/ , pages 16743--16754

    2022

  41. [49]

    , author Maire, M

    author Lin, T.-Y. , author Maire, M. , author Belongie, S. , author Hays, J. , author Perona, P. , author Ramanan, D. , author Doll \'a r, P. , & author Zitnick, C. L. ( year 2014 ). title Microsoft coco: Common objects in context . In booktitle Computer Vision--ECCV 2014: 13t...

    2014

  42. [50]

    , & author Hutter, F

    author Loshchilov, I. , & author Hutter, F. ( year 2017 ). title Decoupled weight decay regularization . journal arXiv preprint arXiv:1711.05101 \/ ,

    2017 arXiv

  43. [52]

    , author Tang, Y

    author Ma, Y. , author Tang, Y. , author Yang, W. , author Zhang, T. , author Zhang, J. , & author Kang, M. ( year 2024 ). title Unifying visual and vision-language tracking via contrastive learning . In booktitle Proceedings of the AAAI Conference on Artificial Intelligence \...

    2024

  44. [53]

    , author Rasheed, H

    author Maaz, M. , author Rasheed, H. , author Khan, S. , & author Khan, F. ( year 2024 ). title Videogpt+: Integrating image and video encoders for enhanced video understanding . journal arXiv preprint arXiv:2406.09418 \/ ,

    2024 arXiv

  45. [54]

    , author Danelljan, M

    author Mayer, C. , author Danelljan, M. , author Bhat, G. , author Paul, M. , author Paudel, D. P. , author Yu, F. , & author Van Gool, L. ( year 2022 ). title Transforming model prediction for tracking . In booktitle Proceedings of the IEEE/CVF conference on computer vision a...

    2022

  46. [55]

    , author Danelljan, M

    author Mayer, C. , author Danelljan, M. , author Paudel, D. P. , & author Van Gool, L. ( year 2021 ). title Learning target candidate association to keep track of what not to track . In booktitle Proceedings of the IEEE/CVF international conference on computer vision \/ (pp. p...

    2021

  47. [56]

    , & author Vondrick, C

    author Menon, S. , & author Vondrick, C. ( year 2023 ). title Visual classification via description from large language models . In booktitle The Eleventh International Conference on Learning Representations \/ . https://openreview.net/forum?id=jlAjNL8z5cs

    2023

  48. [57]

    , author Bibi, A

    author Muller, M. , author Bibi, A. , author Giancola, S. , author Alsubaihi, S. , & author Ghanem, B. ( year 2018 ). title Trackingnet: A large-scale dataset and benchmark for object tracking in the wild . In booktitle Proceedings of the European Conference on Computer Vision...

    2018

  49. [58]

    , & author Tarábek, P

    author Ondrašovič, M. , & author Tarábek, P. ( year 2021 ). title Siamese visual object tracking: A survey . journal IEEE Access \/ , volume 9 \/ , pages 110149--110172 . :10.1109/ACCESS.2021.3101988

    2021

  50. [59]

    , author Gao, J

    author Peng, L. , author Gao, J. , author Liu, X. , author Li, W. , author Dong, S. , author Zhang, Z. , author Fan, H. , & author Zhang, L. ( year 2024 ). title Vasttrack: Vast category visual object tracking . In booktitle The Thirty-eight Conference on Neural Information Pr...

    2024

  51. [60]

    , author Kim, J

    author Radford, A. , author Kim, J. W. , author Hallacy, C. , author Ramesh, A. , author Goh, G. , author Agarwal, S. , author Sastry, G. , author Askell, A. , author Mishkin, P. , author Clark, J. et al. ( year 2021 ). title Learning transferable visual models from natural la...

    2021

  52. [61]

    , author Tsoi, N

    author Rezatofighi, H. , author Tsoi, N. , author Gwak, J. , author Sadeghian, A. , author Reid, I. , & author Savarese, S. ( year 2019 ). title Generalized intersection over union: A metric and a loss for bounding box regression . In booktitle Proceedings of the IEEE/CVF conf...

    2019

  53. [62]

    , author Blattmann, A

    author Rombach, R. , author Blattmann, A. , author Lorenz, D. , author Esser, P. , & author Ommer, B. ( year 2022 ). title High-resolution image synthesis with latent diffusion models . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognit...

    2022

  54. [63]

    , author Kim, J

    author Roth, K. , author Kim, J. M. , author Koepke, A. S. , author Vinyals, O. , author Schmid, C. , & author Akata, Z. ( year 2023 ). title Waffling around for performance: Visual classification with random words and broad concepts . In booktitle Proceedings of the IEEE/CVF ...

    2023

  55. [64]

    author Selvaraju, R. R. , author Cogswell, M. , author Das, A. , author Vedantam, R. , author Parikh, D. , & author Batra, D. ( year 2017 ). title Grad-cam: Visual explanations from deep networks via gradient-based localization . In booktitle 2017 IEEE International Conference...

    2017 doi

  56. [65]

    , author He, S

    author Shao, Y. , author He, S. , author Ye, Q. , author Feng, Y. , author Luo, W. , & author Chen, J. ( year 2024 ). title Context-aware integration of language and visual references for natural language tracking . In booktitle Proceedings of the IEEE/CVF Conference on Comput...

    2024

  57. [66]

    , author Dao, S

    author Shi, H. , author Dao, S. D. , & author Cai, J. ( year 2024 ). title Llmformer: Large language model for open-vocabulary semantic segmentation . journal International Journal of Computer Vision \/ , (pp. pages 1--18 )

    2024

  58. [67]

    , author Yu, J

    author Song, Z. , author Yu, J. , author Chen, Y. P. , & author Yang, W. ( year 2022 ). title Transformer tracking with cyclic shifting window attention . In booktitle 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) \/ (pp. pages 8781--8790 ). :10.11...

    2022

  59. [68]

    , author Yu, F

    author Sun, Y. , author Yu, F. , author Chen, S. , author Zhang, Y. , author Huang, J. , author Li, Y. , author Li, C. , & author Wang, C. ( year 2024 ). title Chattracker: Enhancing visual tracking performance via chatting with multimodal large language model . In editor A. G...

    2024

  60. [69]

    , author Shazeer, N

    author Vaswani, A. , author Shazeer, N. , author Parmar, N. , author Uszkoreit, J. , author Jones, L. , author Gomez, A. N. , author Kaiser, . , & author Polosukhin, I. ( year 2017 ). title Attention is all you need . journal Advances in neural information processing systems \...

    2017

  61. [70]

    , author Zhou, W

    author Wang, N. , author Zhou, W. , author Wang, J. , & author Li, H. ( year 2021 a ). title Transformer meets tracker: Exploiting temporal context for robust visual tracking . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition \/ (p...

    2021

  62. [71]

    , author Shu, X

    author Wang, X. , author Shu, X. , author Zhang, Z. , author Jiang, B. , author Wang, Y. , author Tian, Y. , & author Wu, F. ( year 2021 b ). title Towards more flexible and accurate object tracking with natural language: Algorithms and benchmark . In booktitle IEEE CVPR \/

    2021

  63. [72]

    , author Bai, Y

    author Wei, X. , author Bai, Y. , author Zheng, Y. , author Shi, D. , & author Gong, Y. ( year 2023 ). title Autoregressive visual tracking . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition \/ (pp. pages 9697--9706 )

    2023

  64. [73]

    , author Sun, H

    author Wu, H. , author Sun, H. , author Ji, K. , & author Kuang, G. ( year 2025 ). title Temporal-spatial feature interaction network for multi-drone multi-object tracking . journal IEEE Transactions on Circuits and Systems for Video Technology \/ , volume 35 \/ , pages 1165--...

    2025

  65. [74]

    , author Yang, T

    author Wu, Q. , author Yang, T. , author Liu, Z. , author Wu, B. , author Shan, Y. , & author Chan, A. B. ( year 2023 a ). title Dropmae: Masked autoencoders with spatial-attention dropout for tracking tasks . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vis...

    2023

  66. [75]

    , author Hong, D

    author Wu, X. , author Hong, D. , & author Chanussot, J. ( year 2023 b ). title Uiu-net: U-net in u-net for infrared small object detection . journal IEEE Transactions on Image Processing \/ , volume 32 \/ , pages 364--376 . :10.1109/TIP.2022.3228497

    2023

  67. [77]

    , author Wang, Z

    author Xie, F. , author Wang, Z. , & author Ma, C. ( year 2024 a ). title Diffusiontrack: Point set diffusion model for visual object tracking . In booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) \/ (pp. pages 19113--19124 )

    2024

  68. [78]

    , author Zhong, B

    author Xie, J. , author Zhong, B. , author Mo, Z. , author Zhang, S. , author Shi, L. , author Song, S. , & author Ji, R. ( year 2024 b ). title Autoregressive queries for adaptive tracking with spatio-temporal transformers . In booktitle Proceedings of the IEEE/CVF Conference...

    2024

  69. [79]

    , author Peng, H

    author Yan, B. , author Peng, H. , author Fu, J. , author Wang, D. , & author Lu, H. ( year 2021 ). title Learning spatio-temporal transformer for visual tracking . In booktitle Proceedings of the IEEE/CVF international conference on computer vision \/ (pp. pages 10448--10457 )

    2021

  70. [80]

    , author Kumar, T

    author Yang, Z. , author Kumar, T. , author Chen, T. , author Su, J. , & author Luo, J. ( year 2021 ). title Grounding-tracking-integration . journal IEEE Transactions on Circuits and Systems for Video Technology \/ , volume 31 \/ , pages 3433--3443 . :10.1109/TCSVT.2020.3038720

    2021

  71. [81]

    , author Li, L

    author Yang, Z. , author Li, L. , author Lin, K. , author Wang, J. , author Lin, C.-C. , author Liu, Z. , & author Wang, L. ( year 2023 ). title The dawn of lmms: Preliminary explorations with gpt-4v(ision) . https://arxiv.org/abs/2309.17421. arXiv:2309.17421 http://arxiv.org/...

    2023 arXiv

  72. [82]

    , author Chang, H

    author Ye, B. , author Chang, H. , author Ma, B. , author Shan, S. , & author Chen, X. ( year 2022 ). title Joint feature learning and relation modeling for tracking: A one-stream framework . In booktitle European conference on computer vision \/ (pp. pages 341--357 ). organiz...

    2022

  73. [83]

    , author Xiong, Y

    author Yu, Y. , author Xiong, Y. , author Huang, W. , & author Scott, M. R. ( year 2020 ). title Deformable siamese attention networks for visual object tracking . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition \/ (pp. pages 6728--6737 )

    2020

  74. [84]

    , author Liu, L

    author Zhang, C. , author Liu, L. , author Huang, G. , author Wen, H. , author Zhou, X. , & author Wang, Y. ( year 2024 a ). title Webuot-1m: Advancing deep underwater object tracking with a million-scale benchmark . journal Neurips \/ ,

    2024

  75. [85]

    , author Huang, J

    author Zhang, J. , author Huang, J. , author Jin, S. , & author Lu, S. ( year 2024 b ). title Vision-language models for vision tasks: A survey . journal IEEE Transactions on Pattern Analysis and Machine Intelligence \/ , volume 46 \/ , pages 5625--5644 . :10.1109/TPAMI.2024.3369699

    2024

  76. [86]

    , author Peng, H

    author Zhang, Z. , author Peng, H. , author Fu, J. , author Li, B. , & author Hu, W. ( year 2020 ). title Ocean: Object-aware anchor-free tracking . In editor A. Vedaldi , editor H. Bischof , editor T. Brox , & editor J.-M. Frahm (Eds.), booktitle Computer Vision -- ECCV 2020 ...

    2020

  77. [87]

    , author Wang, D

    author Zhao, H. , author Wang, D. , & author Lu, H. ( year 2023 ). title Representation learning for visual object tracking by masked appearance transfer . In booktitle 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) \/ (pp. pages 18696--18705 ). :10...

    2023

  78. [88]

    , author Zhong, B

    author Zheng, Y. , author Zhong, B. , author Liang, Q. , author Li, G. , author Ji, R. , & author Li, X. ( year 2024 ). title Toward unified token learning for vision-language tracking . journal IEEE Transactions on Circuits and Systems for Video Technology \/ , volume 34 \/ ,...

    2024

  79. [89]

    , author Yang, J

    author Zhou, K. , author Yang, J. , author Loy, C. C. , & author Liu, Z. ( year 2022 a ). title Conditional prompt learning for vision-language models . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition \/ (pp. pages 16816--16825 )

    2022

  80. [90]

    , author Yang, J

    author Zhou, K. , author Yang, J. , author Loy, C. C. , & author Liu, Z. ( year 2022 b ). title Learning to prompt for vision-language models . journal International Journal of Computer Vision \/ , volume 130 \/ , pages 2337--2348

    2022

  81. [91]

    , author Zhou, Z

    author Zhou, L. , author Zhou, Z. , author Mao, K. , & author He, Z. ( year 2023 ). title Joint visual grounding and tracking with natural language specification . In booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition \/ (pp. pages 2315...

    2023

  82. [92]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.