Pith. sign in

REVIEW 4 major objections 4 minor 3 cited by

TAR: Temporal Anchor-Constrained Reasoning for Video Temporal Grounding

T0 review · 4 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Mid-reasoning anchors lift video grounding to 61.1 mIoU

desk verdict The T-anchor reward is a sensible idea and the paper is well written, but the headline claim that the reward drives the gains is not supported by the paper's own ablations. read the letter →

arxiv 2508.07683 v2 pith:WOMC6L6K submitted 2025-08-11 cs.CV cs.AI

Chaohong Guo , Xun Mo , Yongwei Nie , Fei Ma , Xuemiao Xu , Chengjiang Long This is my paper · ORCID
classification cs.CVcs.AI
keywords VideoTemporalGroundingAnchorChain-of-ThoughtReinforcementLearningProgressiveRefinementLargeVision-LanguageModelsSoftIoUBootstrapping
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

TAR (Temporal Anchor-Constrained Reasoning) claims that inserting explicit timestamp anchors as checkpoints inside a vision-language model's chain of thought, and rewarding the model only when later anchors are more accurate than earlier ones, forces the model to progressively refine its temporal predictions while keeping the reasoning grounded in visual evidence. This yields state-of-the-art Charades-STA performance (61.1 mIoU with a 7B model) and, the paper argues, better faithfulness and autonomy than both outcome-only reinforcement learning (Time-R1) and full process supervision (Video-VER). TAR also introduces a self-bootstrapping loop: a standard 7B model under light GRPO generates 30K well-formatted reasoning chains, removing the need for large proprietary teachers. If the mechanism holds, anchor-based reward is a low-cost, auditable middle path for grounding reasoning in time.

What carries the argument

The T-anchor: a timestamp interval enclosed in <T-anchor> tags inside the <think> block, serving as an auditable checkpoint. The driving reward is $r_{\text{TAR}} = \sum_{i=1}^{\hat{s}} i \cdot \text{sIoU}_i + \beta \sum_{i=2}^{\hat{s}} \delta_i - \gamma (\hat{s}-s)^2$, where $\text{sIoU}_i$ is the soft IoU of the $i$-th anchor against the ground truth, $\delta_i = 1$ if $\text{sIoU}_i > \text{sIoU}_{i-1}$ and $-1$ otherwise, and the final term penalizes the anchor count deviating from a target. This couples each intermediate thought to visual evidence and blocks the reward-hacking strategy of emitting infinitely many anchors.

What would settle it

A decisive test: train TAR identically but filter the 30K bootstrapped samples by outcome-only soft IoU, ignoring anchor progress; if Charades-STA mIoU stays near 61.1, the anchor filter is not the cause. Alternatively, shuffle the order of anchors in the reward computation so later anchors receive no accuracy bonus; if performance does not drop, the progressive-refinement signal is not the active ingredient.

Watch

Extended reading notes

Core claim

The paper's central claim is that temporal anchors (T-anchors) act as auditable checkpoints that enforce progressive refinement within a model's reasoning trace. The model first proposes a coarse interval, then, under a reward that penalizes any drop in accuracy, re-grounds its subsequent reasoning in the video and emits a more precise interval, and so on until the final answer. The composite reward combines a format reward, a soft IoU reward (which, unlike standard IoU, gives a penalty signal even when there is no overlap), and a TAR reward consisting of an accuracy-weighted sum of anchor soft IoUs, a binary progressive-refinement term rewarding only accuracy increases, and a quadratic pena

Load-bearing premise

The 30K bootstrapped reasoning chains are selected using the same T-anchor reward (including soft IoU against the ground truth) that is later used in reinforcement learning, so the paper assumes this filtering does not leak the reward's preferences into the model in a way that inflates benchmark numbers beyond genuine temporal grounding.

Editorial extensions

If this is right

  • TAR sets a new Charades-STA record (61.1 mIoU, 50.2 R1@0.7) with a 7B LVLM, surpassing methods that train on additional video datasets.
  • On QVHighlights, TAR exceeds Time-R1 by +6.8 mIoU, +9.9 R1@0.5, and +5.8 R1@0.7, indicating the anchor mechanism transfers across datasets.
  • Zero-shot, TAR achieves 41.1 mIoU on ActivityNet-Captions and 30.6 mIoU on TVGBench, and the progressive-refinement pattern transfers to VQA tasks (MVBench and VideoMME gains).
  • The bootstrapped 30K CoT data are essential: SFT+GRPO reaches 61.1 mIoU and 50.2 R1@0.7, versus 45.9 R1@0.7 for GRPO alone.
  • Ablations show each reward component matters: adding the refinement and count-penalty terms raises R1@0.7 from 31.2 to 50.2.

Reading between the lines

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

  • If the anchor mechanism is the active cause, the same intermediate-checkpoint reward should transfer to any temporally-indexed output—audio grounding, dense video captioning, action segmentation—where the model can guess a value, then be rewarded for refining it against the input.
  • The bootstrapping loop filters 30K samples with the same T-anchor reward later used in RL, so part of the gain over Time-R1 may be a data-selection effect rather than the anchor constraint; a direct test with outcome-filtered CoT would settle this.
  • Table 6 shows soft IoU alone accounts for a large share of the gain (Time-R1(+sIoU) 60.0 vs Time-R1 59.8 on 7B), with the anchor reward adding about +1.1 mIoU; the main value may be a smoother reward landscape, with progressive refinement as a secondary boost.
  • The "textual inertia" argument suggests a broader principle: autoregressive vision-language models need structural interruptions to re-attend to raw visual tokens, predicting that any explicit re-grounding marker could yield similar faithfulness gains.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes TAR (Temporal Anchor-Constrained Reasoning) for video temporal grounding. The method inserts explicitly formatted <T-anchor>...</T-anchor> tags into the model's chain-of-thought, and trains with GRPO using a composite reward: a format reward, a soft IoU reward, and a novel temporal-anchor-constrained reward (weighted anchor accuracy, progressive-refinement incentive, and a length penalty). Because the base 7B model rarely emits well-formed anchors, the authors first run a lightweight GRPO pass on the base model, filter the resulting trajectories to 30K samples using the same reward criteria, perform SFT on this bootstrapped data, and then run GRPO again. Experiments on Charades-STA, QVHighlights, ActivityNet-Captions, and TVGBench show improved mIoU over several RL baselines, and ablations indicate that each reward component contributes to the final score. The paper claims state-of-the-art performance and that the T-anchor mechanism yields faithful, autonomous, and progressively refined reasoning traces.

Significance. If the central causal claim is validated, TAR offers an attractive middle ground between outcome-only RL (Time-R1) and full process supervision (Video-VER): rule-based, auditable intermediate constraints without external reward models or distillation from much larger models. The reward design is clearly specified, the bootstrapping pipeline is cost-effective, and the component ablations in Tables 4–6 are informative. The paper also reports transfer of the learned refinement pattern to VQA tasks. The main reservation is attribution: as reported, the experiments do not cleanly separate the contribution of the T-anchor reward from the contribution of the bootstrapped SFT stage, so the strength of the central claim is currently not fully supported by the evidence.

major comments (4)
  1. [§4.3, Tables 4 and 6] The causal isolation of the T-anchor reward is missing. In Table 4, GRPO-only TAR obtains 59.3 mIoU / 45.9 R1@0.7, while Table 6 reports the Time-R1 7B baseline at 59.8 mIoU / 47.5 R1@0.7 under 'identical data scale.' The full TAR model reaches 61.1 mIoU / 50.2 R1@0.7 only after SFT on the 30K bootstrapped chains, which are themselves filtered using the same T-anchor progressive-refinement criteria (Sec. 3.3). Thus the reported +2.0 mIoU over Time-R1 decomposes into roughly −0.5 mIoU from the anchor reward without SFT and +1.8 mIoU from the SFT stage. To support the abstract's claim that reward constraints on intermediate T-anchors drive the gains, the authors must provide matched ablations in which Time-R1 and Time-R1(+sIoU) also receive the same 30K SFT stage, and ideally a version of TAR without the anchor reward but with the same SFT data. Without this, the improvement could be attri
  2. [§3.2, Eq. (8)] The target anchor count s in the length penalty TARnum(o) = (s^−s)^2 is never specified. This is load-bearing because Table 5 shows TARnum contributes +2.6 R1@0.7 relative to the configuration without it. The authors should state whether s is fixed (e.g., s=3 as in Fig. 2), per-dataset, or dynamically computed, and how deviations are scored when the model outputs fewer or more anchors. Without this, the reward is not fully reproducible and the hyperparameter cannot be assessed.
  3. [§3.3] The bootstrapping filtering pipeline is not concretely specified. The text says each retained trajectory must have a minimum number of <T-anchor> tags and that anchors must show 'progressively increasing localization accuracy against the ground truth, with the specific accuracy thresholds (e.g., sIoU)' but no actual thresholds or minimum counts are given. Given that the 30K SFT set is generated from 186K raw outputs using these criteria, and that this SFT stage is central to the final performance, the exact thresholds (e.g., minimum sIoU values per anchor position, minimum number of anchors, final-reward cutoff) must be reported for reproducibility.
  4. [§4.2, Tables 1 and 6] The Time-R1 7B baseline differs between Table 1 and Table 6. Table 1 lists Time-R1 at mIoU 58.8, R1@0.3 82.8, R1@0.5 72.2, R1@0.7 50.1, while Table 6, under 'identical data scale,' lists Time-R1 at mIoU 59.8, R1@0.3 82.4, R1@0.5 70.6, R1@0.7 47.5. These numbers are inconsistent. Please clarify whether these are different training configurations (e.g., with/without additional data, different sampling, or different random seeds) and ensure all comparisons use the same baseline setting.
minor comments (4)
  1. [§4.2, Table 2] The zero-shot results are selectively highlighted: TAR improves mIoU and R1@0.3 on ActivityNet-Captions and TVGBench, but R1@0.7 is lower than Time-R1 on both benchmarks (19.8 vs 21.4 on ActivityNet; 16.0 vs 16.4 on TVGBench). Please discuss this trade-off explicitly.
  2. [§3.2, Eq. (4)] The soft IoU reward is described as having 'no lower bound,' but its range is not discussed. Since sIoU can be arbitrarily negative for far-away predictions, the unboundedness may interact with reward scaling. A brief remark on clipping or normalization would be helpful.
  3. [§4.3, Fig. 4] The faithfulness/autonomy evaluation using Qwen2.5-VL-72B is only referenced to the supplementary material. Since this is a central qualitative claim, include the scoring prompt and a few scored examples in the main text or clearly state that the full protocol is in the supplement.
  4. [General] Several hyperparameters (α=3, β=1, γ=5) are stated with ablation deferred to the supplementary. Please include the α ablation in the main text or prominently in the supplement, since format reward strength directly affects anchor-format compliance.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the T-anchor reward is defined on ground-truth IoU, and the main gains are benchmarked against external baselines under identical data scale.

full rationale

The paper's central claim is not circular. The T-anchor reward in Eqs. (5)-(9) is computed from ground-truth temporal intervals via soft IoU, not from the model's own outputs, so the training signal is external. The progressive-refinement term δ_i rewards monotonic improvement against ground truth, which is a designed objective rather than a re-description of the model's output. Performance is measured on held-out test sets (e.g., Charades-STA test, QVHighlights val, zero-shot ActivityNet/TVGBench), and Table 6 explicitly states that all compared models are trained on the identical data scale, providing an external control against Time-R1 and Time-R1(+sIoU). The bootstrapped 30K CoT filtering does reuse the same sIoU/progressive criteria as the later reward, creating a self-referential data-selection loop, but this is a training-data construction step and does not make test-time predictions equivalent to the inputs; the model must still generalize to unseen videos. No load-bearing self-citation or uniqueness theorem is invoked. Any request for an additional SFT-matched Time-R1 baseline would be an experimental-completeness concern, not a circularity.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical entities. It relies on tuned reward weights and several unspecified hyperparameters (anchor count, filtering thresholds), which are the main free parameters. The key assumptions are that GRPO optimizes well and that the 72B evaluator is a valid judge of faithfulness.

free parameters (5)
  • alpha (format reward weight) = 3
    Tuned hyperparameter for format reward; ablation said in supplementary.
  • beta (progressive refinement weight) = 1
    Tuned hyperparameter balancing refinement incentive.
  • gamma (length penalty weight) = 5
    Tuned hyperparameter controlling penalty on anchor count deviation.
  • target anchor count s = not specified in main text
    Used in the length penalty TAR_num; the value is not given in the main text and is left to the supplement.
  • sIoU filtering thresholds for bootstrapping = not specified
    The 30K samples are filtered using 'specific accuracy thresholds' that are not quantified in the main text.
assumptions (4)
  • domain assumption Qwen2.5-VL base model can perform temporal reasoning with proper prompting
    The method relies on the base model's ability to generate and then improve T-anchors after SFT; Section 3.3 notes the model initially fails to produce anchors.
  • domain assumption GRPO provides unbiased policy optimization for multimodal RL
    The paper adopts GRPO without justification beyond citing DeepSeek-R1; the optimization landscape for this reward is not analyzed.
  • domain assumption Ground-truth segments used for rewards are correct and unique
    All rewards (sIoU, TAR_sIoU, TAR_refine) are computed against the dataset's single ground-truth segment; multi-annotator variance is ignored.
  • domain assumption The 72B judge's scores reflect true faithfulness and autonomy
    Figure 4 uses Qwen2.5-VL-72B to score reasoning trajectories on a 5-point scale, but no validation of the judge's correlation with human judgments is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TAR: Temporal Anchor-Constrained Reasoning for Video Temporal Grounding." pith.science (2026). https://pith.science/paper/WOMC6L6K

@misc{pith2026250807683,
  author       = {Pith},
  title        = {Pith review of: TAR: Temporal Anchor-Constrained Reasoning for Video Temporal Grounding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WOMC6L6K}},
  note         = {Machine review of arXiv:2508.07683}
}
read the original abstract

Video Temporal Grounding (VTG) aims to localize specific video segments corresponding to natural language queries. While recent Large Vision-Language Models (LVLMs) employ Reinforcement Learning to generate Chains-of-Thought (CoT), they typically rely solely on outcome-based supervision. Consequently, this often leads to hallucinations, where the reasoning process becomes disconnected from the visual content and the final prediction. Existing attempts to mitigate this by relying on external supervision from larger models or separate reward models are computationally expensive and prone to rigid patterns. To address these challenges, we propose TAR (Temporal Anchor-Constrained Reasoning), a framework that introduces the temporal anchor (T-anchor) as a transparent and auditable checkpoint mechanism. T-anchor enforces progressive refinement within the CoT, compelling the model to continuously ground its intermediate thoughts in visual evidence and iteratively calibrate temporal predictions, thereby significantly enhancing the faithfulness and autonomy of the reasoning process and final accuracy. Furthermore, we introduce a bootstrapping paradigm that automatically harvests high-quality CoT data using only a standard 7B model, eliminating the dependency on ultra-large models. Extensive experiments demonstrate that TAR achieves state-of-the-art performance and generates faithful, autonomous, and progressively refined reasoning traces.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

  1. MarkIt: Training-Free Visual Markers for Precise Video Temporal Grounding

    cs.MM 2026-04 unverdicted novelty 7.0 of 10

    MarkIt uses a query-to-mask bridge with open-vocabulary segmentation to add visual markers and frame indices to videos, enabling Vid-LLMs to achieve state-of-the-art temporal grounding on moment retrieval and highligh...

  2. CAVE: Competence-Aware Visual Boundary Evidence Alignment for Video Temporal Grounding

    cs.CL 2026-08 conditional novelty 6.0 of 10

    CAVE adds boundary evidence tokens and an attention alignment reward to RL-based video temporal grounding, improving average timestamp accuracy by about 1.9 points over TimeLens-7B.

  3. TimePLE: Rethinking Temporal Representation for Video Temporal Grounding

    cs.CV 2026-07 conditional novelty 6.0 of 10

    TimePLE predicts a whole video interval as a joint distribution over a position-duration square, rather than predicting start and end separately, and reports higher mIoU across four VTG benchmarks.

Reference graph

Works this paper leans on

43 extracted references · 17 canonical work pages · cited by 3 Pith papers

  1. [1]

    In: Proceedings of the IEEE international conference on computer vision

    Anne Hendricks, L., Wang, O., Shechtman, E., Sivic, J., Darrell, T., Russell, B.: Localizing moments in video with natural language. In: Proceedings of the IEEE international conference on computer vision. pp. 5803–5812 (2017)

    2017

  2. [2]

    In: Proceedings of the ieee conference on computer vision and pattern recognition

    Caba Heilbron, F., Escorcia, V., Ghanem, B., Carlos Niebles, J.: Activitynet: A large-scale video benchmark for human activity understanding. In: Proceedings of the ieee conference on computer vision and pattern recognition. pp. 961–970 (2015)

    2015

  3. [3]

    Cao, Z., Zhang, B., Du, H., Yu, X., Li, X., Wang, S.: Flashvtg: Feature layering and adaptivescorehandlingnetworkforvideotemporalgrounding.In:2025IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). pp. 9226–9236. IEEE (2025)

    work page 2025

  4. [4]

    Datasets and Recipes for Video Temporal Grounding via Reinforcement Learning

    Chen, R., Fan, Z., Luo, T., Zou, H., Feng, Z., Xie, G., Zhang, H., Wang, Z., Liu, Z., Zhang, H.: Datasets and recipes for video temporal grounding via reinforcement learning. arXiv preprint arXiv:2507.18100 (2025)

    work page Pith review arXiv 2025

  5. [5]

    arXiv preprint arXiv:2503.21776 (2025)

    Feng, K., Gong, K., Li, B., Guo, Z., Wang, Y., Peng, T., Wu, J., Zhang, X., Wang, B., Yue, X.: Video-r1: Reinforcing video reasoning in mllms. arXiv preprint arXiv:2503.21776 (2025)

    arXiv 2025

  6. [6]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Fu, C., Dai, Y., Luo, Y., Li, L., Ren, S., Zhang, R., Wang, Z., Zhou, C., Shen, Y., Zhang, M., et al.: Video-mme: The first-ever comprehensive evaluation benchmark of multi-modal llms in video analysis. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 24108–24118 (2025) 16 C. Guo, X. Mo et al

    work page 2025

  7. [7]

    In: Proceedings of the IEEE international conference on computer vision

    Gao, J., Sun, C., Yang, Z., Nevatia, R.: Tall: Temporal activity localization via language query. In: Proceedings of the IEEE international conference on computer vision. pp. 5267–5275 (2017)

    work page 2017

  8. [8]

    In: Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition

    Grauman, K., Westbury, A., Byrne, E., Chavis, Z., Furnari, A., Girdhar, R., Ham- burger, J., Jiang, H., Liu, M., Liu, X., et al.: Ego4d: Around the world in 3,000 hours of egocentric video. In: Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition. pp. 18995–19012 (2022)

    2022

Show all 43 references
  1. [9]

    In: Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition

    Guo, C., He, Y., Nie, Y., Ma, F., Xu, X., Long, C.: T2sgrid: Temporal-to-spatial gridification for video temporal grounding. In: Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition. pp. 3443–3454 (2026)

    2026

  2. [10]

    arXiv preprint arXiv:2501.12948 (2025)

    Guo, D., Yang, D., Zhang, H., Song, J., Zhang, R., Xu, R., Zhu, Q., Ma, S., Wang, P., Bi, X., et al.: Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948 (2025)

    2025 arXiv

  3. [11]

    arXiv preprint arXiv:2410.05643 (2024)

    Guo, Y., Liu, J., Li, M., Liu, Q., Chen, X., Tang, X.: Trace: Temporal grounding video llm via causal event modeling. arXiv preprint arXiv:2410.05643 (2024)

    2024 arXiv

  4. [12]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Huang, B., Wang, X., Chen, H., Song, Z., Zhu, W.: Vtimellm: Empower llm to grasp video moments. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 14271–14280 (2024)

    2024

  5. [13]

    In: Proceedings of the IEEE/CVF Inter- national Conference on Computer Vision

    Jang, J., Park, J., Kim, J., Kwon, H., Sohn, K.: Knowing where to focus: Event- aware transformer for video grounding. In: Proceedings of the IEEE/CVF Inter- national Conference on Computer Vision. pp. 13846–13856 (2023)

    2023

  6. [14]

    arXiv preprint arXiv:2503.09516 (2025)

    Jin, B., Zeng, H., Yue, Z., Yoon, J., Arik, S., Wang, D., Zamani, H., Han, J.: Search-r1: Training llms to reason and leverage search engines with reinforcement learning. arXiv preprint arXiv:2503.09516 (2025)

    2025 arXiv

  7. [15]

    In: Proceedings of the IEEE international conference on computer vision

    Krishna, R., Hata, K., Ren, F., Fei-Fei, L., Carlos Niebles, J.: Dense-captioning events in videos. In: Proceedings of the IEEE international conference on computer vision. pp. 706–715 (2017)

    2017

  8. [16]

    Advances in Neural Information Processing Systems34, 11846–11858 (2021)

    Lei, J., Berg, T.L., Bansal, M.: Detecting moments and highlights in videos via natural language queries. Advances in Neural Information Processing Systems34, 11846–11858 (2021)

    2021

  9. [17]

    arXiv preprint arXiv:2506.01908 (2025)

    Li, H., Han, S., Liao, Y., Luo, J., Gao, J., Yan, S., Liu, S.: Reinforcement learning tuning for videollms: Reward design and data efficiency. arXiv preprint arXiv:2506.01908 (2025)

    2025 arXiv

  10. [18]

    arXiv preprint arXiv:2305.06355 (2023)

    Li, K., He, Y., Wang, Y., Li, Y., Wang, W., Luo, P., Wang, Y., Wang, L., Qiao, Y.: Videochat: Chat-centric video understanding. arXiv preprint arXiv:2305.06355 (2023)

    2023 arXiv

  11. [19]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Li, K., Wang, Y., He, Y., Li, Y., Wang, Y., Liu, Y., Wang, Z., Xu, J., Chen, G., Luo, P., et al.: Mvbench: A comprehensive multi-modal video understanding benchmark. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 22195–22206 (2024)

    2024

  12. [20]

    arXiv preprint arXiv:2501.00574 (2024)

    Li, X., Wang, Y., Yu, J., Zeng, X., Zhu, Y., Huang, H., Gao, J., Li, K., He, Y., Wang, C., et al.: Videochat-flash: Hierarchical compression for long-context video modeling. arXiv preprint arXiv:2501.00574 (2024)

    2024 arXiv

  13. [21]

    arXiv preprint arXiv:2504.06958 (2025)

    Li,X.,Yan,Z.,Meng,D.,Dong,L.,Zeng,X.,He,Y.,Wang,Y.,Qiao,Y.,Wang,Y., Wang, L.: Videochat-r1: Enhancing spatio-temporal perception via reinforcement fine-tuning. arXiv preprint arXiv:2504.06958 (2025)

    2025 arXiv

  14. [22]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Lin, K.Q., Zhang, P., Chen, J., Pramanick, S., Gao, D., Wang, A.J., Yan, R., Shou, M.Z.: Univtg: Towards unified video-language temporal grounding. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 2794–2804 (2023) TAR 17

    2023

  15. [23]

    arXiv preprint arXiv:2510.06077 (2025)

    Luo, M., Xue, Z., Dimakis, A., Grauman, K.: When thinking drifts: Evidential grounding for robust video reasoning. arXiv preprint arXiv:2510.06077 (2025)

    2025

  16. [24]

    arXiv preprint arXiv:2306.05424 (2023)

    Maaz, M., Rasheed, H., Khan, S., Khan, F.S.: Video-chatgpt: Towards de- tailed video understanding via large vision and language models. arXiv preprint arXiv:2306.05424 (2023)

    2023 arXiv

  17. [25]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Miech, A., Zhukov, D., Alayrac, J.B., Tapaswi, M., Laptev, I., Sivic, J.: Howto100m: Learning a text-video embedding by watching hundred million nar- rated video clips. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 2630–2640 (2019)

    2019

  18. [26]

    CoRR (2023)

    Moon, W., Hyun, S., Lee, S.B., Heo, J.P.: Correlation-guided query-dependency calibration in video representation learning for temporal grounding. CoRR (2023)

    2023

  19. [27]

    In: ICASSP 2021- 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

    Oncescu, A.M., Henriques, J.F., Liu, Y., Zisserman, A., Albanie, S.: Queryd: A video dataset with high-quality text and audio narrations. In: ICASSP 2021- 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). pp. 2265–2269. IEEE (2021)

    2021

  20. [28]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Qu,M.,Chen,X.,Liu,W.,Li,A.,Zhao,Y.:Chatvtg:Videotemporalgroundingvia chat with video dialogue large language models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 1847–1856 (2024)

    2024

  21. [29]

    Regneri, M., Rohrbach, M., Wetzel, D., Thater, S., Schiele, B., Pinkal, M.: Ground- ingactiondescriptionsinvideos.TransactionsoftheAssociationforComputational Linguistics1, 25–36 (2013)

    2013

  22. [30]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Ren, S., Yao, L., Li, S., Sun, X., Hou, L.: Timechat: A time-sensitive multi- modal large language model for long video understanding. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 14313– 14323 (2024)

    2024

  23. [31]

    arXiv preprint arXiv:2505.07263 (2025)

    Wang, X., Wang, P., Pei, J., Shen, W., Peng, Y., Hao, Y., Qiu, W., Jian, A., Xie, T., Song, X., et al.: Skywork-vl reward: An effective reward model for multimodal understanding and reasoning. arXiv preprint arXiv:2505.07263 (2025)

    2025 arXiv

  24. [32]

    arXiv preprint arXiv:2503.13377 (2025)

    Wang, Y., Wang, Z., Xu, B., Du, Y., Lin, K., Xiao, Z., Yue, Z., Ju, J., Zhang, L., Yang, D., et al.: Time-r1: Post-training large vision language model for temporal video grounding. arXiv preprint arXiv:2503.13377 (2025)

    2025 arXiv

  25. [33]

    arXiv preprint arXiv:2307.06942 (2023)

    Wang, Y., He, Y., Li, Y., Li, K., Yu, J., Ma, X., Li, X., Chen, G., Chen, X., Wang, Y., et al.: Internvid: A large-scale video-text dataset for multimodal understanding and generation. arXiv preprint arXiv:2307.06942 (2023)

    2023 arXiv

  26. [34]

    arXiv preprint arXiv:2403.10228 (2024)

    Wang, Y., Meng, X., Liang, J., Wang, Y., Liu, Q., Zhao, D.: Hawkeye: Train- ing video-text llms for grounding text in videos. arXiv preprint arXiv:2403.10228 (2024)

    2024 arXiv

  27. [35]

    arXiv preprint arXiv:2506.20670 (2025)

    Wu, J., Deng, Z., Li, W., Liu, Y., You, B., Li, B., Ma, Z., Liu, Z.: Mmsearch-r1: Incentivizing lmms to search. arXiv preprint arXiv:2506.20670 (2025)

    2025 arXiv

  28. [36]

    5: Visual test-time scaling to reinforce multimodal reasoning by iterative perception

    Yan, Z., Li, X., He, Y., Yue, Z., Zeng, X., Wang, Y., Qiao, Y., Wang, L., Wang, Y.: Videochat-r1. 5: Visual test-time scaling to reinforce multimodal reasoning by iterative perception. arXiv preprint arXiv:2509.21100 (2025)

    2025

  29. [37]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Yang, A., Nagrani, A., Seo, P.H., Miech, A., Pont-Tuset, J., Laptev, I., Sivic, J., Schmid, C.: Vid2seq: Large-scale pretraining of a visual language model for dense video captioning. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10...

    2023

  30. [38]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Zala, A., Cho, J., Kottur, S., Chen, X., Oguz, B., Mehdad, Y., Bansal, M.: Hierarchical video-moment retrieval and step-captioning. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 23056– 23065 (2023) 18 C. Guo, X. Mo et al

    2023

  31. [39]

    In: The Thirteenth International Con- ference on Learning Representations (2025),https://openreview.net/forum?id= nAVejJURqZ

    Zeng, X., Li, K., Wang, C., Li, X., Jiang, T., Yan, Z., Li, S., Shi, Y., Yue, Z., Wang, Y., Wang, Y., Qiao, Y., Wang, L.: Timesuite: Improving MLLMs for long video understanding via grounded tuning. In: The Thirteenth International Con- ference on Learning Representations (202...

    2025

  32. [40]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2021)

    Zhang, S., Peng, H., Fu, J., Lu, Y., Luo, J.: Multi-scale 2d temporal adjacency networks for moment localization with natural language. IEEE Transactions on Pattern Analysis and Machine Intelligence (2021)

    2021

  33. [41]

    In: Proceedings of the AAAI Confer- ence on Artificial Intelligence (2020)

    Zhang, S., Peng, H., Fu, J., Luo, J.: Learning 2d temporal adjacent networks for moment localization with natural language. In: Proceedings of the AAAI Confer- ence on Artificial Intelligence (2020)

    2020

  34. [42]

    arXiv preprint arXiv:2505.15436 (2025)

    Zhang, X., Gao, Z., Zhang, B., Li, P., Zhang, X., Liu, Y., Yuan, T., Wu, Y., Jia, Y., Zhu, S.C., et al.: Chain-of-focus: Adaptive visual search and zooming for multimodal reasoning via rl. arXiv preprint arXiv:2505.15436 (2025)

    2025 arXiv

  35. [43]

    arXiv preprint arXiv:2508.11630 (2025)

    Zhang, Y.F., Lu, X., Yin, S., Fu, C., Chen, W., Hu, X., Wen, B., Jiang, K., Liu, C., Zhang, T., et al.: Thyme: Think beyond images. arXiv preprint arXiv:2508.11630 (2025)

    2025 arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.