Pith. sign in

REVIEW 4 major objections 6 minor 43 references

ATSTrack: Enhancing Visual-Language Tracking by Aligning Temporal and Spatial Scales

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Mismatched temporal and spatial scales, not missing cross-modal interaction, is what limits visual-language tracking.

desk verdict Strong TNL2K result and a reasonable architectural extension, but the undocumented LLM decomposition and internal contradictions make this a conditional accept rather than a clean one. read the letter →

arxiv 2507.00454 v1 pith:L2AS4FRG submitted 2025-07-01 cs.CV cs.AI

Yihao Zhen , Qiang Wang , Yu Qiao , Liangqiong Qu , Huijie Fan This is my paper · ORCID
classification cs.CVcs.AI
keywords visual-languagetrackingsingleobjecttemporalandspatialscalealignmentfine-grainedfeaturemodulationtokenmultimodaltransformernaturallanguagespecification
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

The paper tries to establish that the remaining bottleneck in visual-language tracking is not a lack of cross-modal interaction but a mismatch in temporal and spatial scales between language phrases and visual evidence. It claims that decomposing each language description into four attributes, category, appearance, action, and location, and modifying each attribute's features with the visual input that shares its scale improves tracking. It also claims that feeding a visual-language token from the previous frame into the visual backbone guides feature extraction toward language-relevant content. If right, this would make natural-language descriptions a reliably stronger tracking cue, especially in long-term and adversarially described sequences, and it would explain the reported 66.2 AUC on TNL2K, 3.1 points above the prior best language tracker.

What carries the argument

The central mechanism is the four-way attribute decomposition together with the Fine-Grained Modification (FGM) module and the Visual-Language (VL) token. The four attributes, category, appearance, action, and location, are defined by which visual input matches their temporal and spatial scale; FGM then applies a custom operation per attribute, with the category feature used to suppress background in the latest template, action features cross-attending over the whole template sequence, and location features gated against the search frame. The VL token, formed by concatenating the average of the modified language features with the visual class token, is fed into the visual backbone of the next frame to bias feature extraction toward language-relevant content.

What would settle it

A direct test would be to take the trained ATSTrack and replace the LLM attribute parsing with a fixed randomized assignment of phrases to the four attributes, or with one unified attribute, while keeping everything else identical, then measure AUC on TNL2K; if the score does not drop below the reported 66.2 by a margin comparable to the FGM ablation gain of about 1.4 AUC on LaSOT, then the fine-grained alignment itself is not doing the work.

Watch

Extended reading notes

Core claim

On its own terms, the central discovery is that explicit alignment of the temporal and spatial granularity of language and visual inputs is what makes feature modification work in visual-language tracking. Rather than treating all language features and all visual features as two wholes, ATSTrack decomposes each description into category, appearance, action, and location phrases, then routes each phrase to a matching visual source: category and appearance to the latest template, action to the whole template sequence, and location to the search frame. The Fine-Grained Modification module applies a different operation per attribute, including a visual feature modification that suppresses background in the template using category features and a language feature ablation that gates out location tokens unrelated to the search image. A visual-language token assembled from the previous frame's modified language features and the visual class token is injected into the next frame's backbone, so visual features are extracted under linguistic guidance. The reported consequence is state-of-the-art or competitive results on TNL2K, LaSOT, and OTBlang.

Load-bearing premise

The whole method depends on a language model cleanly splitting each sentence into exactly four parts, category, appearance, action, and location, and if that split is noisy, the fine-grained alignment gains vanish and the model degrades to ordinary cross-attention.

Editorial extensions

If this is right

  • On TNL2K, ATSTrack reaches 66.2 AUC, 3.1 points above the previous best language tracker and 3.5 points above the best visual tracker, suggesting scale alignment specifically helps adversarial and camera-switch samples.
  • On LaSOT, the 72.6 AUC is competitive with top visual-only trackers and 1.8 points above the second-best language tracker, indicating a benefit for long-term tracking.
  • Ablations on LaSOT attribute roughly 1.4 AUC to the fine-grained modification module and 0.6 AUC to the visual-language token, so both components contribute to the overall gain.
  • Attribute-level ablations show appearance descriptions matter most, while action and location descriptions contribute less in the current datasets.
  • Concatenating visual and language information into a single VL token outperforms cross-attention between the two tokens in the reported experiments.

Reading between the lines

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

  • The fixed four-attribute decomposition is a heuristic tied to how existing datasets phrase descriptions; datasets with richer relational or temporal language would likely need additional attribute types or a learned decomposer for the benefit to transfer.
  • If the LLM segmentation is noisy, part of the FGM gain could come from an implicit regularization effect rather than from true scale alignment; replacing LLM parsing with human-annotated attribute labels could separate these two explanations.
  • The VL-token mechanism is a lightweight way to propagate language through a visual backbone, and a similar token could be applied to other vision-language tasks such as referring expression segmentation or video grounding, where spatial-scale mismatch also occurs.
  • The finding that ungated location descriptions can interfere suggests future datasets should annotate location relative to changing scene context, and the LFA module may partly compensate for annotation bias.
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. The paper proposes ATSTrack, a vision-language tracker that addresses temporal and spatial scale misalignment between visual and language inputs. Language descriptions are decomposed by an unspecified LLM into four attributes (category, appearance, action, location), each routed to different visual inputs in a Fine-Grained Modification module: category suppresses background in the template, action attends over a template sequence, location is gated against the search region, and appearance is refined by the modified template. A Visual-Language token, formed by concatenating the average of modified language features with the visual CLS token, is fed into the visual backbone of the next frame. Experiments on TNL2K, LaSOT, and OTBlang report an AUC of 66.2 on TNL2K (3.1 points above prior language trackers), 72.6 on LaSOT, and 71.0 on OTBlang (second best). Ablations on LaSOT are presented for FGM components, gating weight α, VL-token components, and attribute removal. The stated contributions claim state-of-the-art performance on three tracking datasets.

Significance. If the results hold, ATSTrack would be a strong empirical contribution to vision-language tracking: the TNL2K improvement (66.2 AUC vs. 63.1 for UVLTrack-B) is substantial, and the LaSOT number is competitive with strong visual-only trackers. The conceptual framing of attribute-level temporal/spatial correspondence is appealing and leads to a modular architecture (FGM, VL token) that is easy to understand. The paper also includes several useful sanity ablations (attribute removal, token modality removal) and states that code will be released. The main significance is contingent on reproducibility of the LLM-based attribute decomposition and on the soundness of the ablation/hyperparameter choices, as detailed below.

major comments (4)
  1. [§4.1 and §3.2–3.3] The central mechanism depends on decomposing each language description into four attributes, but the LLM used for this decomposition is never specified. §4.1 states only that 'we use LLM to segment the language descriptions into different attributes,' with no model name, prompt, temperature, or post-processing, and no evaluation of segmentation quality. This omission is load-bearing because the FGM routing, the LFA gating, and the VL token all operate on this decomposition. If the LLM misassigns phrases, the reported gains from fine-grained alignment cannot be attributed to the proposed temporal/spatial alignment mechanism. The authors should specify the LLM and prompt, provide segmentation examples with failure cases, and ideally report a human-evaluated accuracy on a sample of the TNL2K and LaSOT training descriptions.
  2. [§4.3, Table 2c] The gating weight α is selected on the LaSOT benchmark (α=50 is reported in §3.3, while Table 2c's caption states the best value is α=25, and the text says 'LFA achieves the best performance with α = 25'), and the same LaSOT benchmark is then used for the final state-of-the-art comparison in Table 1. This is test-set fitting in a mild form: the final LaSOT AUC (72.6) is reported after choosing a hyperparameter on the same benchmark. The authors should either (a) report a validation split or cross-validation procedure for choosing α, (b) use a different benchmark or validation set for hyperparameter selection, or (c) at minimum show that the final LaSOT result is stable across a range of α values and report the sensitivity honestly.
  3. [§4.3, Table 2c and text] There is a direct internal contradiction: Table 2c reports the best AUC at α=50 (72.0), but the text states that 'LFA achieves the best performance with α = 25.' Moreover, §3.3 sets α=50 in the method description, which is consistent with Table 2c but inconsistent with the text. This contradiction must be resolved: the authors must correct the text or the table and state unambiguously which α is used in the final model.
  4. [§4.3, Table 2b and text] The ablation narrative for the Visual-Language token contradicts its own table. The text says 'Without using the Visual-language token (w/o token), the model decreases in the AUC score by 0.6%,' which matches the 72.0→72.6 improvement. However, it then says 'Using the Language token independently (w/o V token) leads to a decrease in the AUC score by 0.3%,' while Table 2b lists w/o V token at 71.7, which is a 0.9-point drop from the full model (72.6), not 0.3. Table 2b also shows w/o L token at 72.0 (0.6 below full), while the text says 'Using the visual token independently (w/o L token) does not lead to notable changes.' The text and table should be made consistent, and the conclusions about the contribution of each token should be re-derived from the correct numbers.
minor comments (6)
  1. [§3.3, Eq. for LFA] The gating formula uses a scaling factor α with α=50 in the main text, but the ablation section reports α=25 and α=50 inconsistencies as noted in the major comments; the notation should be consistent and the chosen value should be stated once in the method.
  2. [Table 1 caption] The caption says 'LaSOText' but the table columns list LaSOT and OTBlang; the dataset names should be corrected for consistency.
  3. [§3.1, Figure 2] The overview text refers to 'TSATrack' in §3.1 ('Fig. 2 shows the general framework of the TSATrack'), which should be 'ATSTrack'.
  4. [§4.2] The LaSOT paragraph claims ATSTrack 'outperforms the second best visual-language tracker by 1.8% in term of AUC,' but the table shows the second best VL tracker on LaSOT AUC is DecoupleTNL at 71.2, giving a gap of 1.4 points, not 1.8; this should be corrected or clarified which metric is meant.
  5. [§4.3, Table 3] The 'Effect of Each attribute' section describes results, but the table rows are not labeled with the corresponding AUC in the text in a way that allows a reader to trace each claimed drop; I recommend adding the numeric deltas directly in the text or in the table.
  6. [References] There are several reference formatting errors (e.g., [9] author list 'Vitaly Feng Qi and, Ablavsky, Qinxun Bai', [15] and [21] missing venue page details); the authors should proofread the bibliography.

Circularity Check

1 steps flagged · score 2.0 of 10

No derivation-level circularity; the only circularity-adjacent issue is that the LFA gating weight is selected on LaSOT and then LaSOT is reused for the headline comparison.

  1. fitted input called prediction [Section 3.3 (LFA equation, alpha=50), Section 4.3 (Table 2c), and Table 1 (LaSOT results)]
    "Where α = 50, φ = 0.5. ... We analyze the impact of different gating weights α on the effect of different on LFA performance on LaSOT. As shown in Tab. 2c, LFA achieves the best performance with α = 25."

    The LFA gating scale α is selected by a hyperparameter sweep reported on LaSOT (Table 2c), and the same LaSOT benchmark is then used to report the final ATSTrack result in Table 1. Thus the LaSOT comparison is not an independent, out-of-sample test of the LFA design: the scalar α was chosen to maximize exactly the benchmark whose AUC is subsequently advertised. This is a mild evaluation circularity (test-set hyperparameter selection), not a derivation-level equivalence: the architecture's behavior on TNL2K and OTBlang is not determined by this sweep, and the α choice does not by itself force the reported gains.

full rationale

The paper contains no mathematical derivation that is equivalent to its inputs by construction. The FGM equations, LFA gating formula, and VL-token aggregation are explicit module definitions, not hidden restatements of the final AUC numbers. There is no load-bearing self-citation chain: the cited prior work is external, and no uniqueness theorem is imported from the authors' own papers. The unspecified LLM used for attribute segmentation is a reproducibility and robustness gap rather than a circular step, since the reported gains could in principle fail if the segmentation is noisy, but the paper does not define the segmentation in terms of the output. The only concrete circularity-adjacent practice is selecting the LFA weight α on LaSOT (Table 2c) and then reusing LaSOT for the headline comparison (Table 1), which weakens the independence of the LaSOT claim but does not invalidate the TNL2K or OTBlang comparisons. I therefore find no significant derivation-level circularity.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim relies on the assumed reliability of LLM-based attribute decomposition and on hyperparameters tuned on the evaluation benchmark. No new physical or conceptual entities are introduced.

free parameters (3)
  • α (LFA gating scale) = 50 (main results), 25 claimed best in text
    Controls the sharpness of the gating sigmoid in Language Feature Ablation. Tuned on LaSOT in Table 2c, with the text contradicting the table.
  • φ (LFA threshold spread) = 0.5
    Weights the standard deviation in the threshold θ = mid + φ·std. Set by hand with no ablation.
  • λ1, λ2 (loss weights) = 5, 2
    Standard loss weights adopted from prior work, no ablation shown.
assumptions (3)
  • domain assumption Language descriptions can be reliably decomposed into exactly four attributes (category, appearance, action, location).
    Section 3.2 states this decomposition is performed by an unspecified LLM. If the segmentation is noisy, the whole framework degrades.
  • domain assumption Each attribute corresponds best to one type of visual input (appearance to latest template, action to template sequence, location to search region).
    Section 3.2 asserts these correspondences without empirical justification beyond the ablations.
  • domain assumption Averaging the modified language features preserves the information needed to guide visual extraction in the next frame.
    Section 3.4 uses global average pooling to form Tlang, which may discard spatial structure.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ATSTrack: Enhancing Visual-Language Tracking by Aligning Temporal and Spatial Scales." pith.science (2026). https://pith.science/paper/L2AS4FRG

@misc{pith2026250700454,
  author       = {Pith},
  title        = {Pith review of: ATSTrack: Enhancing Visual-Language Tracking by Aligning Temporal and Spatial Scales},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/L2AS4FRG}},
  note         = {Machine review of arXiv:2507.00454}
}
read the original abstract

A main challenge of Visual-Language Tracking (VLT) is the misalignment between visual inputs and language descriptions caused by target movement. Previous trackers have explored many effective feature modification methods to preserve more aligned features. However, an important yet unexplored factor ultimately hinders their capability, which is the inherent differences in the temporal and spatial scale of information between visual and language inputs. To address this issue, we propose a novel visual-language tracker that enhances the effect of feature modification by \textbf{A}ligning \textbf{T}emporal and \textbf{S}patial scale of different input components, named as \textbf{ATSTrack}. Specifically, we decompose each language description into phrases with different attributes based on their temporal and spatial correspondence with visual inputs, and modify their features in a fine-grained manner. Moreover, we introduce a Visual-Language token that comprises modified linguistic information from the previous frame to guide the model to extract visual features that are more relevant to language description, thereby reducing the impact caused by the differences in spatial scale. Experimental results show that our proposed ATSTrack achieves performance comparable to existing methods. Our code will be released.

Figures

Figures reproduced from arXiv: 2507.00454 by the authors.

Figure 1
Figure 1. Comparison with other Visual-Language trackers. (a) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed ATSTrack framework. ATSTrack has been improved in two aspects: 1) A Visual-language token is [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. (a) The structure of the visual feature modification mod [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: AUC score of different attributes in LaSOT. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: The attention map of Visual-Language token. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Visualized results of the proposed ATSTrack on three challenging scenarios for visual object tracking: severe occlusion, fast [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 41 canonical work pages

  1. [1]

    Hiptrack: Visual tracking with historical prompts

    Wenrui Cai, Qingjie Liu, and Yunhong Wang. Hiptrack: Visual tracking with historical prompts. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 19258–19267, 2023. 2

    work page 2024

  2. [2]

    Robust object modeling for visual tracking

    Yidong Cai, Jie Liu, Jie Tang, and Gangshan Wu. Robust object modeling for visual tracking. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 9589–9600, 2023. 2

    work page 2023

  3. [3]

    Ost: Refining text knowledge with optimal spatio-temporal descriptor for general video recognition

    Tom Tongjia Chen, Hongshan Yu, Zhengeng Yang, Zechuan Li, Wei Sun, and Chen Chen. Ost: Refining text knowledge with optimal spatio-temporal descriptor for general video recognition. 2024 IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition (CVPR), pages 18888–18898,

    work page 2024

  4. [4]

    Seqtrack: Sequence to sequence learning for visual ob- ject tracking

    Xin Chen, Houwen Peng, Dong Wang, Huchuan Lu, and Han Hu. Seqtrack: Sequence to sequence learning for visual ob- ject tracking. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 14572– 14581, 2023. 2, 6

    work page 2023

  5. [5]

    Mixformer: End-to-end tracking with iterative mixed atten- tion

    Yutao Cui, Cheng Jiang, Limin Wang, and Gangshan Wu. Mixformer: End-to-end tracking with iterative mixed atten- tion. In Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition , pages 13608–13618,

    work page

  6. [6]

    MixFormerV2: Efficient Fully Transformer Tracking

    Yutao Cui, Tian-Shu Song, Gangshan Wu, and Liming Wang. Mixformerv2: Efficient fully transformer tracking. ArXiv, abs/2305.15896, 2023. 6

    work page Pith review arXiv 2023

  7. [7]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020. 5

    arXiv 2010

  8. [8]

    Lasot: A high-quality benchmark for large-scale single ob- ject tracking

    Heng Fan, Liting Lin, Fan Yang, Peng Chu, Ge Deng, Sijia Yu, Hexin Bai, Yong Xu, Chunyuan Liao, and Haibin Ling. Lasot: A high-quality benchmark for large-scale single ob- ject tracking. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 5369–5378,

    work page 2019

Show all 43 references
  1. [9]

    Siamese natural language tracker: Tracking by nat- ural language descriptions with siamese trackers

    Vitaly Feng Qi and, Ablavsky, Qinxun Bai, and Stan Sclaroff. Siamese natural language tracker: Tracking by nat- ural language descriptions with siamese trackers. In CVPR,

  2. [10]

    Generalized relation modeling for transformer tracking

    Shenyuan Gao, Chunluan Zhou, and Jun Zhang. Generalized relation modeling for transformer tracking. 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 18686–18695, 2023. 5

    2023

  3. [11]

    Siamcar: Siamese fully convolutional classification and regression for visual tracking

    Dongyan Guo, Jun Wang, Ying Cui, Zhenhua Wang, and Shengyong Chen. Siamcar: Siamese fully convolutional classification and regression for visual tracking. In Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6269–6277, 2020. 2

    2020

  4. [12]

    Divert more attention to vision-language tracking

    Mingzhe Guo, Zhipeng Zhang, Heng Fan, and Liping Jing. Divert more attention to vision-language tracking. In Pro- ceedings of the 36th International Conference on Neural In- formation Processing Systems, Red Hook, NY , USA, 2022. Curran Associates Inc. 3, 6

    2022

  5. [13]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll´ar, and Ross Girshick. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 16000– 16009, 2022. 5

    2022

  6. [14]

    Target-aware tracking with long-term context at- tention

    Kaijie He, Canlong Zhang, Sheng Xie, Zhixin Li, and Zhi- wen Wang. Target-aware tracking with long-term context at- tention. In AAAI Conference on Artificial Intelligence, 2023. 2

    2023

  7. [15]

    A multi-modal global in- stance tracking benchmark (mgit): better locating target in complex spatio-temporal and causal relationship

    Shiyu Hu, Dailing Zhang, Meiqi Wu, Xiaokun Feng, Xuchen Li, Xin Zhao, and Kaiqi Huang. A multi-modal global in- stance tracking benchmark (mgit): better locating target in complex spatio-temporal and causal relationship. In Pro- ceedings of the 37th International Conference on...

    2023

  8. [16]

    Got-10k: A large high-diversity benchmark for generic object tracking in the wild

    Lianghua Huang, Xin Zhao, and Kaiqi Huang. Got-10k: A large high-diversity benchmark for generic object tracking in the wild. IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, 43(5):1562–1577, 2021. 6

    2021

  9. [17]

    Rtracker: Recoverable tracking via pn tree structured memory

    Yuqing Huang, Xin Li, Zikun Zhou, Yaowei Wang, Zhenyu He, and Ming-Hsuan Yang. Rtracker: Recoverable tracking via pn tree structured memory. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 19038–19047, 2024. 2

    2024

  10. [18]

    Towards sequence-level training for vi- sual tracking

    Minji Kim, Seungkwang Lee, Jungseul Ok, Bohyung Han, and Minsu Cho. Towards sequence-level training for vi- sual tracking. In European Conference on Computer Vision,

  11. [19]

    Citetracker: Correlating image and text for visual tracking

    Xin Li, Yuqing Huang, Zhenyu He, Yaowei Wang, Huchuan Lu, and Ming-Hsuan Yang. Citetracker: Correlating image and text for visual tracking. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 9974– 9983, 2023. 2, 3, 6, 8

    2023

  12. [20]

    Dtllm-vlt: Diverse text generation for visual language tracking based on llm

    Xuchen Li, Xiaokun Feng, Shiyu Hu, Meiqi Wu, Dailing Zhang, Jing Zhang, and Kaiqi Huang. Dtllm-vlt: Diverse text generation for visual language tracking based on llm. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops , pages ...

    2024

  13. [21]

    Beyond mot: Semantic multi- object tracking

    Yunhao Li, Hao Wang, Xue Ma, Jiali Yao, Shaohua Dong, Heng Fan, and Libo Zhang. Beyond mot: Semantic multi- object tracking. ArXiv, abs/2403.05021, 2024. 1

    2024 arXiv

  14. [22]

    Zhenyang Li, Ran Tao, Efstratios Gavves, Cees G. M. Snoek, and Arnold W. M. Smeulders. Tracking by natural language specification. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 7350–7358, 2017. 3

    2017

  15. [23]

    Swintrack: A simple and strong baseline for trans- former tracking

    Liting Lin, Heng Fan, Zhipeng Zhang, Yong Xu, and Haibin Ling. Swintrack: A simple and strong baseline for trans- former tracking. In Advances in Neural Information Process- ing Systems , pages 16743–16754. Curran Associates, Inc.,

  16. [24]

    Tracking meets lora: Faster training, larger model, stronger performance

    Liting Lin, Heng Fan, Zhipeng Zhang, Yaowei Wang, Yong Xu, and Haibin Ling. Tracking meets lora: Faster training, larger model, stronger performance. In European Confer- ence on Computer Vision, 2024. 6

    2024

  17. [25]

    Tracking by natural language specification with long short-term context decoupling

    Ding Ma and Xiangqian Wu. Tracking by natural language specification with long short-term context decoupling. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 14012–14021, 2023. 2, 3, 6

    2023

  18. [26]

    Unifying visual and vision-language tracking via contrastive learning

    Yinchao Ma, Yuyang Tang, Wenfei Yang, Tianzhu Zhang, Jinpeng Zhang, and Mengxue Kang. Unifying visual and vision-language tracking via contrastive learning. In Pro- ceedings of the AAAI Conference on Artificial Intelligence , pages 4107–4116, 2024. 2, 6

    2024

  19. [27]

    Trackingnet: A large-scale dataset and benchmark for object tracking in the wild

    Matthias Muller, Adel Bibi, Silvio Giancola, Salman Al- subaihi, and Bernard Ghanem. Trackingnet: A large-scale dataset and benchmark for object tracking in the wild. In Proceedings of the European conference on computer vision (ECCV), pages 300–317, 2018. 6

    2018

  20. [28]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, ...

    2021

  21. [29]

    Context-aware integration of lan- guage and visual references for natural language tracking

    Yanyan Shao, Shuting He, Qi Ye, Yuchao Feng, Wenhan Luo, and Jiming Chen. Context-aware integration of lan- guage and visual references for natural language tracking. In 2024 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 19208–19217, 2024. 1, 2...

    2024

  22. [30]

    Chat- tracker: Enhancing visual tracking performance via chatting with multimodal large language model

    Yiming Sun, Fan Yu, Shaoxiang Chen, Yu Zhang, Junwei Huang, Yang Li, Chenhui Li, and Changbo Wang. Chat- tracker: Enhancing visual tracking performance via chatting with multimodal large language model. In The Thirty-eighth Annual Conference on Neural Information Processing Sy...

    2024

  23. [31]

    Towards more flexible and accurate object tracking with natural language: Algo- rithms and benchmark

    Xiao Wang, Xiujun Shu, Zhipeng Zhang, Bo Jiang, Yaowei Wang, Yonghong Tian, and Feng Wu. Towards more flexible and accurate object tracking with natural language: Algo- rithms and benchmark. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition ,...

    2021

  24. [32]

    Autoregressive visual tracking

    Xing Wei, Yifan Bai, Yongchao Zheng, Dahu Shi, and Yihong Gong. Autoregressive visual tracking. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 9697–9706, 2023. 2, 6

    2023

  25. [33]

    Improving visual grounding with multi-scale discrep- ancy information and centralized-transformer

    Jie Wu, Chunlei Wu, Fuyan Wang, Leiquan Wang, and Yiwei Wei. Improving visual grounding with multi-scale discrep- ancy information and centralized-transformer. Expert Sys- tems with Applications, 247:123223, 2024. 2

    2024

  26. [34]

    Qiangqiang Wu, Tianyu Yang, Ziquan Liu, Baoyuan Wu, Ying Shan, and Antoni B. Chan. Dropmae: Masked au- toencoders with spatial-attention dropout for tracking tasks. 2023 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 14561–14571, 2023. 6

    2023

  27. [35]

    Object track- ing benchmark

    Yi Wu, Jongwoo Lim, and Ming-Hsuan Yang. Object track- ing benchmark. IEEE Transactions on Pattern Analysis and Machine Intelligence, 37(9):1834–1848, 2015. 6

    2015

  28. [36]

    Correlation-aware deep tracking

    Fei Xie, Chunyu Wang, Guangting Wang, Yue Cao, Wankou Yang, and Wenjun Zeng. Correlation-aware deep tracking. 2022 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 8741–8750, 2022. 2

    2022

  29. [37]

    Autore- gressive queries for adaptive tracking with spatio-temporal transformers

    Jinxia Xie, Bineng Zhong, Zhiyi Mo, Shengping Zhang, Liangtao Shi, Shuxiang Song, and Rongrong Ji. Autore- gressive queries for adaptive tracking with spatio-temporal transformers. 2024 IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition (CVPR), pages 19300–19309,

    2024

  30. [38]

    Learning spatio-temporal transformer for vi- sual tracking

    Bin Yan, Houwen Peng, Jianlong Fu, Dong Wang, and Huchuan Lu. Learning spatio-temporal transformer for vi- sual tracking. In Proceedings of the IEEE/CVF international conference on computer vision, pages 10448–10457, 2021. 2

    2021

  31. [39]

    Joint feature learning and relation modeling for tracking: A one-stream framework

    Botao Ye, Hong Chang, Bingpeng Ma, Shiguang Shan, and Xilin Chen. Joint feature learning and relation modeling for tracking: A one-stream framework. In European Conference on Computer Vision, pages 341–357. Springer, 2022. 2, 5, 6, 8

    2022

  32. [40]

    Exploring the feature extraction and relation modeling for light-weight transformer tracking

    Jikai Zheng, Mingjiang Liang, Shaoli Huang, and Jifeng Ning. Exploring the feature extraction and relation modeling for light-weight transformer tracking. In Computer Vision – ECCV 2024, pages 110–126, Cham, 2025. Springer Nature Switzerland. 5

    2024

  33. [41]

    Ji, and Xianxian Li

    Yaozong Zheng, Bineng Zhong, Qihua Liang, Guorong Li, R. Ji, and Xianxian Li. Toward unified token learning for vision-language tracking. IEEE Transactions on Circuits and Systems for Video Technology, 34:2125–2135, 2023. 6

    2023

  34. [42]

    Odtrack: Online dense temporal token learning for visual tracking

    Yaozong Zheng, Bineng Zhong, Qihua Liang, Zhiyi Mo, Shengping Zhang, and Xianxian Li. Odtrack: Online dense temporal token learning for visual tracking. In Proceed- ings of the AAAI conference on artificial intelligence , pages 7588–7596, 2024. 6

    2024

  35. [43]

    Joint visual grounding and tracking with natural language specifi- cation

    Li Zhou, Zikun Zhou, Kaige Mao, and Zhenyu He. Joint visual grounding and tracking with natural language specifi- cation. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 23151–23160, 2023. 1, 2, 6

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.