Pith. sign in

REVIEW 5 major objections 3 minor 3 cited by

LeAdQA: LLM-Driven Context-Aware Temporal Grounding for Video Question Answering

T0 review · 5 major / 3 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 LeAdQA rewrites each answer option into a causal scene description, then grounds the video to it before answering.

desk verdict A plausible grounding-based VideoQA recipe with consistent gains, but the key ablation conflates grounding with rewriting and the NExT-QA comparison is val-vs-test; referee-worthy but needs major revision. read the letter →

arxiv 2507.14784 v2 pith:YE6SKFD7 submitted 2025-07-20 cs.CV cs.AI

Xinxin Dong , Baoyun Peng , Haokai Ma , Yufei Wang , Zixuan Dong , Fei Hu , Xiaodong Wang This is my paper
classification cs.CVcs.AI
keywords VideoQuestionAnsweringtemporalgroundingcausalreasoningLLM-drivenrewritingmultiple-choiceQAlongunderstandingmultimodallargelanguagemodelsNExT-QA
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 argues that a major bottleneck in VideoQA is not model capacity but deciding which frames to watch. Its central move is to rewrite each question-option pair into a causal event description—a hypothetical video situation in which that option is true—and use those rewritten queries to drive a temporal grounding model that localizes the relevant video segments. The grounded segments, fused across options, are then passed to a multimodal LLM for the final answer. On NExT-QA, IntentQA, and NExT-GQA, the pipeline improves accuracy over the same MLLMs used alone, and grounded sampling reaches uniform-sampling accuracy with fewer frames. If correct, this reframes VideoQA as an evidence-retrieval problem rather than a whole-video comprehension problem.

What carries the argument

The load-bearing component is the LLM-driven question-option rephrasing: a structured prompt asks GPT-4o to convert each pair $(Q, o_i)$ into a causal-enriched description of a video situation in which $o_i$ would be true. This turns answer selection into a set of retrieval queries, so the temporal grounding stage (foreground, boundary, and saliency heads over clip features, trained with contrastive and IoU losses) can localize evidence instead of relying on uniform sampling. A two-level interval fusion (intra-option then inter-option, with an IoU threshold of 0.3) then consolidates the evidence before the MLLM reads the sampled frames.

What would settle it

Hold out a fresh set of videos and questions written in the same style as NExT-QA that GPT-4o has not seen, run LeAdQA with and without the GPT-4o rewriting, and check whether the causal-rewriting gain persists; alternatively, replace GPT-4o with a paraphrase model trained only on synthetic data and compare. If the gain disappears on unseen questions, the reported improvement comes from answer leakage rather than from grounding.

Watch

Extended reading notes

Core claim

LeAdQA treats each answer option as a candidate claim about the video and asks an LLM to spell out the situation that would make that claim true. The rephrased query $q'_i = M_r(P_r(Q, o_i))$ is then fed into a motion-aware temporal grounding model that predicts, per clip, a foreground flag, boundary offsets, and a saliency score. Candidate intervals are merged within and across options via an IoU-based fusion, and the surviving segments supply keyframes to an MLLM that generates the answer. The paper reports that this combination raises Tarsier-34B from 75.2 to 77.0 on NExT-GQA QA accuracy, from 79.3 to 80.6 on NExT-QA, and from 76.9 to 78.5 on IntentQA; the ablation attributes +1.3 points on NExT-QA to the causal rewriting itself, and shows QA accuracy improving as grounding quality improves (random 78.0, uniform 79.3, ground-truth segments 82.1).

Load-bearing premise

That rewriting each option with GPT-4o on public benchmarks does not leak the correct answer; if GPT-4o already knows these benchmark questions, the measured grounding and accuracy gains may be inflated.

Editorial extensions

If this is right

  • With LeAdQA, VideoQA accuracy becomes a function of temporal grounding quality: the paper shows accuracy climbing from 78.0 (random frames) to 79.3 (uniform sampling) to 82.1 (ground-truth segments) on NExT-QA, so improving grounding should transfer directly to QA accuracy.
  • Query-focused sampling within grounded intervals reaches 81.2% accuracy at 32 frames where uniform sampling needs 48, making the method a frame-efficiency lever for long-video MLLMs.
  • The gains are not tied to a single backbone: both Qwen2.5-VL and Tarsier models improve when LeAdQA is added, suggesting the rephrasing-plus-grounding recipe is model-agnostic.
  • Causal rewriting helps most on causal questions, including a +4.2 gain on Causal-How questions in IntentQA, indicating that rewriting supplies event-chain structure that the visuals alone do not provide.

Reading between the lines

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

  • If the leak-free assumption holds, the same rephrasing trick could be applied to other multiple-choice visual tasks, such as image QA, embodied QA, or video caption ranking, where options can be converted into hypothetical scene descriptions.
  • The paper's own numbers suggest an upper bound: if grounding were perfect, accuracy would rise considerably further (82.1 vs. 80.6 with causal rewriting), so investing in grounding quality may yield larger returns than scaling the MLLM.
  • One can test the causal-rewriting hypothesis directly by training a grounding model on original question-option pairs and on rewritten pairs with the same videos and comparing temporal IoU; the paper reports that higher tIoU correlates with QA accuracy but does not isolate grounding quality with and without rewriting.
  • Treating answer selection as retrieval opens a natural extension: feeding multiple fused intervals back into the MLLM with their timestamps so the model can reason about event order, not just about which content is present.
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

5 major / 3 minor

Summary. The paper proposes LeAdQA, a VideoQA framework that combines LLM-driven rewriting of question-option pairs with a temporal grounding module (UniVTG) and an interval-fusion mechanism, then feeds the selected segments to an MLLM for answer generation. Experiments on NExT-GQA, NExT-QA, and IntentQA report consistent accuracy gains over base MLLMs, and an ablation shows that ground-truth segments outperform uniform sampling, which in turn outperforms random sampling. The central claim is that causal question-option rewriting improves temporal grounding and thereby improves VideoQA accuracy.

Significance. If the central claim is upheld, LeAdQA is a practically attractive, training-efficient wrapper that can be added to existing MLLMs such as Tarsier and Qwen2.5-VL, with reported gains of 0.5--1.6 points on three benchmarks. The paper explicitly credits the grounding benefit with a clean ordering across random, uniform, and ground-truth sampling, and it tests multiple MLLM backbones and frame budgets. However, the attribution of the gains to the causal-rewriting mechanism is not isolated by the current ablations, and the main NExT-QA comparison mixes validation and test splits. These issues are fixable but are load-bearing for the paper's headline claims.

major comments (5)
  1. [Results and Analysis, Table 3] The text states that LeAdQA is evaluated on the test sets of NExT-GQA and IntentQA and on the validation set of NExT-QA, yet Table 3 compares LeAdQA against baselines whose reported numbers in the cited works are test-set accuracies. Comparing a validation-set number with published test-set numbers is not apples-to-apples and can inflate the reported SOTA claim. Please either evaluate on the official NExT-QA test split or re-run all baselines on the same validation split, and clearly label the split in the table and caption.
  2. [Ablation Study, Table 5] The 'Varying Rewriting (w/ Uniform Grounding)' block does not isolate causal rewriting. The row 'w/o Causal Rewriting' reports 79.3, identical to the 'Uniform Sampling' row in the 'Varying Grounding' block. Since the enriched queries q'_i are used only to produce grounded segments and are not included in the final MLLM prompt (Equations 14--17), toggling rewriting while keeping uniform sampling should change nothing; the comparison therefore conflates 'adding grounding' with 'adding causal rewriting'. The missing control is grounding with the original question-option tuples q_i=(Q,o_i) as UniVTG queries, with all other components identical. A second useful control is a non-causal paraphrase of the question-option pairs, to separate the effect of causal rewriting from the effect of converting options into natural-language sentences.
  3. [Table 1] The LeAdQA-34B row in Table 1 reports only Acc@QA (77.0) and Acc@GQA (20.7); all grounding metrics (mIoP, IoP@0.3, IoP@0.5, mIoU, IoU@0.3, IoU@0.5) are missing for this strongest model. Since the paper's central claim is that temporal grounding drives the QA gains, the absence of grounding metrics for the best-performing configuration prevents the reader from verifying the grounding benefit at the largest scale. Please report the full grounding metric set for LeAdQA-34B.
  4. [In-depth Analysis and Tables 3--5] No error bars, multiple-seed runs, or significance tests are reported anywhere. Several headline gains are small (for example +0.5 on NExT-QA for Tarsier-7B and +0.8 on IntentQA for Tarsier-7B in Table 4), and the interval-fusion hyperparameters (Top-K, IoU threshold) appear to be tuned on the NExT-QA validation set. Without variance estimates or a significance test, it is difficult to judge whether these differences are robust. Please provide at least three seeds or a paired significance test for the main comparisons and ablations.
  5. [Method, Eq. (1), and Ablation Study, Table 5] There is a correctness-risk concern about leakage from GPT-4o rewriting. The prompt Pr asks GPT-4o to write a hypothetical video situation under which option o_i is true; for the correct option this description is likely to match the actual video, while for incorrect options it is not. Because NExT-QA, IntentQA, and NExT-GQA are public benchmarks, GPT-4o may have prior knowledge of the answers, which could make the grounded segments systematically favor the correct option and inflate the measured grounding and QA improvements. A concrete test would be to compare against a neutral rephrasing instruction that does not assume the truth of the option (e.g., 'rewrite the question-option pair while preserving its original semantics') and to measure grounding quality with the original q_i as the query. This would separate genuine reasoning improvement from answer leakage via the LLM's parametric memory.
minor comments (3)
  1. [Experimental Settings, Datasets] The dataset name 'IntenQA' appears to be a typo for 'IntentQA', which is the name used elsewhere in the paper, including Table 4.
  2. [Related Works, Video Question Answering] The sentence 'VideoAgent (Wang et al. 2025) futher employs LLMs as iterative information extractors' contains the typo 'futher'; it should be 'further'.
  3. [Implementation Details and Table 2] Table 2 labels the base models as 'Qwen-3B' and 'Qwen-7B', while the Implementation Details section says the authors use 'Qwen2.5-VL-3B-Instruct and Qwen2.5-VL-7B-Instruct'. Please unify the naming to avoid ambiguity about which Qwen version is evaluated.

Circularity Check

1 steps flagged · score 6.0 of 10

Causal-rewriting ablation reduces to a no-grounding versus grounding+rewriting comparison; headline SOTA results are otherwise self-contained.

  1. self definitional [Ablation Study, Table 5; method Eqs. (1) and (14)-(17)]
    "Varying Grounding (w/ Causal Rewriting) ... Uniform Sampling 74.4 80.5 85.3 79.3 ... Varying Rewriting (w/ Uniform Grounding) ... w/o Causal Rewriting 74.4 80.5 85.3 79.3 ... + Causal Rewriting (ours) 75.7 81.9 86.8 80.6"

    The 'w/o Causal Rewriting' row is numerically identical to the Uniform Sampling baseline (no grounding). Because the enriched queries q'_i are only consumed by the grounding module and are not part of the final answering prompt Pa(Q,O) in Eqs. (14)-(17), disabling rewriting while keeping uniform sampling changes nothing in the MLLM input. The reported +1.3 point gain (79.3 to 80.6) is therefore the joint effect of adding grounding plus rewriting, not the isolated effect of causal rewriting. The missing control row is grounding with the original (Q,o_i) queries, so the paper's attribution of the gain to 'causal rewriting' reduces by construction to a no-grounding versus grounding+rewriting contrast.

full rationale

The headline SOTA comparisons in Tables 1-4 compare LeAdQA against external baselines on public benchmarks, so those accuracy gains are not circular. The grounding benefit itself is independently supported in Table 5 (random 78.0 < uniform 79.3 < ground-truth segments 82.1). The only load-bearing step that reduces by construction is the causal-rewriting attribution: the 'w/o Causal Rewriting' control is identical to the uniform-sampling/no-grounding row, so the reported +1.3 points cannot be assigned to rewriting alone. Two further validity concerns, distinct from circularity, are the GPT-4o leakage risk on public benchmarks (Eq. 1) and the tuning of NExT-QA thresholds on the same validation split used for its headline number; these affect interpretability but are not circular derivations under the stated rules.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central claim is an engineering claim, not a derivation. The ledger records the empirical dependencies: GPT-4o rewriting quality, UniVTG grounding quality, MLLM answer reliability, and the thresholds selected on validation data.

free parameters (5)
  • saliency threshold tau_p = not specified
    Filters out low-relevance grounded segments; stated to be tuned on a validation set.
  • Top-K candidate intervals = 5 (optimal)
    Number of grounded segments kept per query; evaluated over {1,3,5} on NExT-QA validation, with Top-5 best in Table 6.
  • IoU merge threshold tau_m = 0.3
    Intervals with IoU above this threshold are merged; grid [0.1,0.3,0.5,0.7,0.9] evaluated, 0.3 chosen as optimal in Table 6.
  • number of keyframes Kf = 8 for 7B, 48 for 34B
    Frames sampled from grounded segments for the MLLM; evaluated over [1,2,4,8,16,32,48] and selected per model scale in Figure 3.
  • loss weights lambda_L1, lambda_IoU, lambda_inter, lambda_intra = unspecified
    Weights in the grounding losses (Eq. 7-8); values are not reported.
assumptions (3)
  • domain assumption GPT-4o can produce causally enriched query descriptions q'_i for each option without leaking the answer.
    The entire grounding pipeline depends on the quality and neutrality of these rewrites; introduced in Section 'LLM-Driven Question-Option Rephrase', Eq. 1.
  • domain assumption UniVTG provides reliable temporal grounding on NExT-QA, IntentQA, and NExT-GQA without task-specific fine-tuning.
    The method adopts UniVTG as the grounding model; no fine-tuning details are given.
  • domain assumption The MLLMs (Tarsier, Qwen2.5-VL) can correctly answer using only keyframes from grounded segments.
    The answer decoder is used as-is, with only a trainable projection MLP; training details for that projection are omitted.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LeAdQA: LLM-Driven Context-Aware Temporal Grounding for Video Question Answering." pith.science (2026). https://pith.science/paper/YE6SKFD7

@misc{pith2026250714784,
  author       = {Pith},
  title        = {Pith review of: LeAdQA: LLM-Driven Context-Aware Temporal Grounding for Video Question Answering},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YE6SKFD7}},
  note         = {Machine review of arXiv:2507.14784}
}
read the original abstract

Video Question Answering (VideoQA) requires identifying sparse critical moments in long videos and reasoning about their causal relationships to answer semantically complex questions. While recent advances in multimodal learning have improved alignment and fusion, current approaches remain limited by two prevalent but fundamentally flawed strategies: (1) task-agnostic sampling indiscriminately processes all frames, overwhelming key events with irrelevant content; and (2) heuristic retrieval captures superficial patterns but misses causal-temporal structures needed for complex reasoning. To address these challenges, we introduce LeAdQA, an innovative approach that bridges these gaps through synergizing causal-aware query refinement with fine-grained visual grounding. Our method first leverages LLMs to reformulate question-option pairs, resolving causal ambiguities and sharpening temporal focus. These refined queries subsequently direct a temporal grounding model to precisely retrieve the most salient segments, complemented by an adaptive fusion mechanism dynamically integrating the evidence to maximize relevance. The integrated visual-textual cues are then processed by an MLLM to generate accurate, contextually-grounded answers. Experiments on NExT-QA, IntentQA, and NExT-GQA demonstrate that our method's precise visual grounding substantially enhances the understanding of video-question relationships, achieving state-of-the-art (SOTA) performance on complex reasoning tasks while maintaining computational efficiency.

Figures

Figures reproduced from arXiv: 2507.14784 by the authors.

Figure 1
Figure 1. Architecture comparison: (a) Traditional frame [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The architecture of LeAdQA. Question-option pairs are first rephrased by LLMs to generate enhanced descriptions, [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Tarsier-7B (left) and Tarsier-34B (right): VideoQA [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 3 Pith papers

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

  1. Counterfactual Reasoning for Fine-Grained Evidence Disentanglement in VideoQA

    cs.CV 2026-06 unverdicted novelty 6.0 of 10

    CREDiT applies counterfactual reasoning via structural causal models to decompose video representations into causal and non-causal parts for more reliable VideoQA on datasets like NExT-GQA and SportsQA.

  2. VTI-CoT: Visual-Textual Interleaved Chain of Thought for Video Reasoning

    cs.CV 2026-06 unverdicted novelty 6.0 of 10

    VTI-CoT proposes a visual-textual interleaved chain-of-thought method for video reasoning, built via automated annotation and OCR compression, claiming SOTA performance and better training efficiency on same-scale models.

  3. UpstreamQA: A Modular Framework for Explicit Reasoning on Video Question Answering Tasks

    cs.CV 2026-04 unverdicted novelty 5.0 of 10

    UpstreamQA disentangles video reasoning by using LRMs for explicit upstream object identification and scene context before downstream LMM VideoQA, improving performance and interpretability on OpenEQA and NExTQA in so...

Reference graph

Works this paper leans on

45 extracted references · 28 canonical work pages · cited by 3 Pith papers

  1. [1]

    Achiam, S

    J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023

    arXiv 2023

  2. [2]

    Alayrac, J

    J.-B. Alayrac, J. Donahue, P. Luc, A. Miech, I. Barr, Y. Hasson, K. Lenc, A. Mensch, K. Millican, M. Reynolds, et al. Flamingo: a visual language model for few-shot learning. Advances in neural information processing systems, 35: 0 23716--23736, 2022

    2022

  3. [3]

    W. Chu, H. Xue, Z. Zhao, D. Cai, and C. Yao. The forgettable-watcher model for video question answering. Neurocomputing, 314: 0 386--393, 2018

    work page 2018

  4. [4]

    H. Fei, S. Wu, W. Ji, H. Zhang, M. Zhang, M.-L. Lee, and W. Hsu. Video-of-thought: Step-by-step video reasoning from perception to cognition. In Forty-first International Conference on Machine Learning, 2024

    work page 2024

  5. [5]

    T.-J. Fu, L. Li, Z. Gan, K. Lin, W. Y. Wang, L. Wang, and Z. Liu. An empirical study of end-to-end video-language transformers with masked visual modeling. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 22898--22909, 2023

    work page 2023

  6. [6]

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

    work page 2017

  7. [7]

    M. M. Islam, N. Ho, X. Yang, T. Nagarajan, L. Torresani, and G. Bertasius. Video recap: Recursive captioning of hour-long videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18198--18208, 2024

    work page 2024

  8. [8]

    Kahatapitiya, K

    K. Kahatapitiya, K. Ranasinghe, J. Park, and M. S. Ryoo. Language repository for long video understanding. arXiv preprint arXiv:2403.14622, 2024

    arXiv 2024

Show all 45 references
  1. [9]

    W. Kim, C. Choi, W. Lee, and W. Rhee. An image grid can be worth a video: Zero-shot video question answering using a vlm. arXiv preprint arXiv:2403.18406, 2024

    2024 arXiv

  2. [10]

    J. Lei, L. Yu, M. Bansal, and T. L. Berg. Tvqa: Localized, compositional video question answering. arXiv preprint arXiv:1809.01696, 2018

    2018 arXiv

  3. [11]

    J. Lei, T. L. Berg, and M. Bansal. Detecting moments and highlights in videos via natural language queries. Advances in Neural Information Processing Systems, 34: 0 11846--11858, 2021

    2021

  4. [12]

    J. Li, D. Li, S. Savarese, and S. Hoi. BLIP-2 : Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models , June 2023 a

    2023

  5. [13]

    J. Li, P. Wei, W. Han, and L. Fan. Intentqa: Context-aware video intent reasoning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 11963--11974, 2023 b

    2023

  6. [14]

    Y. Li, X. Wang, J. Xiao, W. Ji, and T.-S. Chua. Invariant grounding for video question answering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2928--2937, 2022

    2022

  7. [15]

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

    2023

  8. [16]

    H. Liu, C. Li, Q. Wu, and Y. J. Lee. Visual instruction tuning. Advances in neural information processing systems, 36, 2024

    2024

  9. [17]

    M. Liu, X. Wang, L. Nie, Q. Tian, B. Chen, and T.-S. Chua. Cross-modal moment localization in videos. In Proceedings of the 26th ACM international conference on Multimedia, pages 843--851, 2018

    2018

  10. [18]

    J. Lu, J. Yang, D. Batra, and D. Parikh. Hierarchical question-image co-attention for visual question answering. Advances in neural information processing systems, 29, 2016

    2016

  11. [19]

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

    2023 arXiv

  12. [20]

    J. Min, S. Buch, A. Nagrani, M. Cho, and C. Schmid. Morevqa: Exploring modular reasoning models for video question answering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13235--13245, 2024

    2024

  13. [21]

    W. Moon, S. Hyun, S. Park, D. Park, and J.-P. Heo. Query-dependent video representation for moment retrieval and highlight detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 23023--23033, 2023

    2023

  14. [22]

    J. Mun, M. Cho, and B. Han. Local-global video-text interactions for temporal grounding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10810--10819, 2020

    2020

  15. [23]

    J. Park, K. Ranasinghe, K. Kahatapitiya, W. Ryoo, D. Kim, and M. S. Ryoo. Too many frames, not all useful: Efficient strategies for long-form video qa. arXiv preprint arXiv:2406.09396, 2024

    2024

  16. [24]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, et al. Learning transferable visual models from natural language supervision. In International conference on machine learning, pages 8748--8763. PMLR, 2021

    2021

  17. [25]

    Ranasinghe, X

    K. Ranasinghe, X. Li, K. Kahatapitiya, and M. S. Ryoo. Understanding long videos in one multimodal language model pass. arXiv preprint arXiv:2403.16998, 2024

    2024 arXiv

  18. [26]

    S. Ren, K. He, R. Girshick, and J. Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. IEEE transactions on pattern analysis and machine intelligence, 39 0 (6): 0 1137--1149, 2016

    2016

  19. [27]

    Seo, G.-C

    A. Seo, G.-C. Kang, J. Park, and B.-T. Zhang. Attend what you need: Motion-appearance synergistic networks for video question answering. arXiv preprint arXiv:2106.10446, 2021

    2021 arXiv

  20. [28]

    R. Tan, X. Sun, P. Hu, J.-h. Wang, H. Deilamsalehy, B. A. Plummer, B. Russell, and K. Saenko. Koala: Key frame-conditioned long video-llm. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13581--13591, 2024

    2024

  21. [29]

    D. Tran, L. Bourdev, R. Fergus, L. Torresani, and M. Paluri. Learning spatiotemporal features with 3d convolutional networks. In Proceedings of the IEEE international conference on computer vision, pages 4489--4497, 2015

    2015

  22. [30]

    J. Wang, L. Yuan, Y. Zhang, and H. Sun. Tarsier: Recipes for training and evaluating large video description models. URL https://arxiv. org/abs/2407.00634, 8, 2024 a

    2024 arXiv

  23. [31]

    L. Wang, Y. Xiong, Z. Wang, Y. Qiao, D. Lin, X. Tang, and L. Van Gool. Temporal segment networks: Towards good practices for deep action recognition. In European conference on computer vision, pages 20--36. Springer, 2016

    2016

  24. [32]

    P. Wang, A. Yang, R. Men, J. Lin, S. Bai, Z. Li, J. Ma, C. Zhou, J. Zhou, and H. Yang. Ofa: Unifying architectures, tasks, and modalities through a simple sequence-to-sequence learning framework. In International conference on machine learning, pages 23318--23340. PMLR, 2022 a

    2022

  25. [33]

    X. Wang, Y. Zhang, O. Zohar, and S. Yeung-Levy. Videoagent: Long-form video understanding with large language model as agent. In European Conference on Computer Vision, pages 58--76. Springer, 2025

    2025

  26. [34]

    Y. Wang, K. Li, Y. Li, Y. He, B. Huang, Z. Zhao, H. Zhang, J. Xu, Y. Liu, Z. Wang, et al. Internvideo: General video foundation models via generative and discriminative learning. arXiv preprint arXiv:2212.03191, 2022 b

    2022 arXiv

  27. [35]

    Z. Wang, S. Yu, E. Stengel-Eskin, J. Yoon, F. Cheng, G. Bertasius, and M. Bansal. Videotree: Adaptive tree-based video representation for llm reasoning on long videos. arXiv preprint arXiv:2405.19209, 2024 b

    2024 arXiv

  28. [36]

    J. Xiao, X. Shang, A. Yao, and T.-S. Chua. Next-qa: Next phase of question-answering to explaining temporal actions. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9777--9786, 2021

    2021

  29. [37]

    J. Xiao, A. Yao, Y. Li, and T.-S. Chua. Can i trust your answer? visually grounded video question answering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13204--13214, 2024

    2024

  30. [38]

    H. Xu, K. He, L. Sigal, S. Sclaroff, and K. Saenko. Text-to-clip video retrieval with early fusion and re-captioning. arXiv preprint arXiv:1804.05113, 2 0 (6): 0 7, 2018

    2018 arXiv

  31. [39]

    Y. Xu, Y. Wei, S. Zhong, X. Chen, J. Qi, and B. Wu. Exploring question guidance and answer calibration for visually grounded video question answering. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 3121--3133, 2024

    2024

  32. [40]

    A. Yang, A. Miech, J. Sivic, I. Laptev, and C. Schmid. Zero-shot video question answering via frozen bidirectional language models. Advances in Neural Information Processing Systems, 35: 0 124--141, 2022

    2022

  33. [41]

    S. Yu, J. Cho, P. Yadav, and M. Bansal. Self-chained image-language model for video localization and question answering. Advances in Neural Information Processing Systems, 36, 2024

    2024

  34. [42]

    T. Yu, J. Yu, Z. Yu, Q. Huang, and Q. Tian. Long-term video question answering via multimodal hierarchical memory attentive networks. IEEE Transactions on Circuits and Systems for Video Technology, 31 0 (3): 0 931--944, 2020

    2020

  35. [43]

    Zhang, T

    C. Zhang, T. Lu, M. M. Islam, Z. Wang, S. Yu, M. Bansal, and G. Bertasius. A simple llm framework for long-range video question-answering. arXiv preprint arXiv:2312.17235, 2023

    2023 arXiv

  36. [44]

    Zhang, H

    S. Zhang, H. Peng, J. Fu, and J. Luo. Learning 2d temporal adjacent networks for moment localization with natural language. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 12870--12877, 2020

    2020

  37. [45]

    a henb \

    Y. Zhao, I. Misra, P. Kr \"a henb \"u hl, and R. Girdhar. Learning video representations from large language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 6586--6597, 2023

    2023

Pith tools

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