Pith. sign in

REVIEW 5 major objections 6 minor 1 cited by

VideoMind: An Omni-Modal Video Dataset with Intent Grounding for Deep-Cognitive Video Understanding

T0 review · 5 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 VideoMind is a 103K-video dataset that annotates every clip with factual, abstract, and intent descriptions, and shows that today's video models retrieve only about 35% of intent-level queries versus about 80% for factual ones.

desk verdict A genuinely useful omni-modal video dataset with a new intent layer, but the intent ground truth is not yet established, so the headline retrieval gap should be read as provisional. read the letter →

arxiv 2507.18552 v1 pith:4ZAVOTUO submitted 2025-07-24 cs.CV cs.AI

Baoyao Yang , Wanyun Li , Dixin Chen , Junxiang Chen , Wenbin Yao , Haifeng Lin This is my paper
classification cs.CVcs.AI
keywords videounderstandingintentgroundingomni-modaldatasetcross-modalretrievalchain-of-thoughtmultimodallargelanguagemodeldeep-cognitivebenchmark
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

VideoMind is a new dataset of 103,000 video clips, each paired with audio and a detailed written description organized in three layers: what is factually seen and heard, an abstract summary, and the intent behind the video—why the uploader posted it and why the person in it acts that way. The paper's core claim is that this intent layer, produced by step-by-step Chain-of-Thought prompting, makes VideoMind the first dataset that supports deep-cognitive video understanding rather than surface captioning. On a 3,000-sample human-validated benchmark, the paper reports that leading video foundation models retrieve videos using intent-level text with only about 35% Rank-1 accuracy, compared with about 80% for factual text. That gap is the paper's evidence that current models know what happens in a video but not why it happens. If the annotations are trustworthy, VideoMind provides a training and evaluation resource for making video models reason about purpose and motivation.

What carries the argument

The load-bearing mechanism is a three-stage Chain-of-Thought generation pipeline run on a multimodal large language model. Stage one produces five separate factual descriptions covering visual content, background audio, OCR, ASR, and text; stage two synthesizes an abstract summary; stage three infers intent through two role-playing tasks (uploader and protagonist) using the fixed template '[subject] aims to [intent] by [action].' Each intent is then double-validated: its key terms are compared by embedding similarity against another mLLM's independent speculation, and experts assess a text-to-video reconstruction to judge whether the intent is plausible. The evaluation protocol—separate retrieval with factual, abstract, and intent queries, plus 'hit any layer' and 'hit all layer' video-to-text metrics—is what surfaces the depth gap.

What would settle it

Have independent human annotators write intent expressions for a random subset of the 3,000 benchmark videos without seeing the model-generated labels, then run text-to-video retrieval using those human intents as queries. If the human-annotated queries retrieve the correct videos at a much higher rate than the model-generated ones (or much lower), the model-generated intent labels do not correspond to human-level understanding of intent.

Watch

Extended reading notes

Core claim

The paper introduces VideoMind and claims it is the first deep-cognitive omni-modal video dataset. Each of its 103K samples decomposes the video into five non-overlapping factual channels—visual, non-speech audio, OCR, ASR, and raw text—then summarizes those into an abstract layer, and finally speculates on intent using the fixed expression '[subject] aims to [intent] by [action],' generated separately from the uploader's and the protagonist's perspectives. A held-out set of 3,000 samples is manually validated as a gold-standard benchmark. In hybrid-cognitive retrieval experiments, the authors find that five widely used video foundation models achieve about 80% Rank-1 retrieval with factual queries but only about 35% with intent queries, and that retrieving all three layers of ground-truth text within the top 10 is rare, with the lowest-ranked layer often beyond position 500. The paper interprets this drop as evidence that these models lack the deep understanding needed to represent video intent.

Load-bearing premise

The whole argument rests on the intent labels being genuine ground truth for the uploader's and protagonist's purposes, even though they are generated by a model and validated only through similarity to another model's output and experts' judgment of a generated video.

Editorial extensions

If this is right

  • If the intent layer is valid, training on VideoMind could give video foundation models a learning signal that aligns representations with purpose and motivation, improving intent recognition, emotion recognition, and content moderation.
  • The benchmark provides a standardized test that existing video datasets lack, allowing future models to be compared on deep understanding instead of surface description.
  • The fixed '[subject] aims to [intent] by [action]' format may allow automatic extraction of subject, action, and intent tags, supporting fine-grained recognition tasks.
  • The ~35% intent retrieval rate can serve as a baseline; any model that surpasses it on the VideoMind-3K benchmark would demonstrate a measurable improvement in deep video understanding.

Reading between the lines

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

  • One testable extension the paper does not run: fine-tune a video foundation model on the 100K training split and measure intent retrieval on the 3K test set; if performance does not improve, the dataset still evaluates but does not teach intent grounding.
  • The dual-validation design conflates intent with plausibility: a generated video can look reasonable even when the stated intent is wrong, so the benchmark may be more reliable for retrieval than for ground-truth intent semantics.
  • The observed gap between factual and intent retrieval might be partially explained by distribution shift in text styles, not only by semantic depth; an ablation that matches factual texts for length and format would separate the two.
  • If other groups apply the same COT pipeline to different base mLLMs and produce different intent labels, the benchmark's stability across generators becomes an important test of whether it measures video intent or model priors.
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 / 6 minor

Summary. VideoMind is introduced as a 103K-sample omni-modal video dataset in which each clip is paired with audio and a three-layer textual description (factual, abstract, intent) generated by the mLLM Qwen2.5-Omni in a chain-of-thought procedure, supplemented by 6W tags and a 3,000-sample manually validated benchmark. The paper reports hybrid-cognitive text-to-video and video-to-text retrieval results for several video foundation models, finding that intent-layer queries achieve markedly lower R@1 (~35%) than factual-layer queries (~80%), which the authors interpret as evidence that current models lack deep video understanding. The dataset and benchmark are publicly released.

Significance. If the intent-layer annotations are valid ground truth, VideoMind would be a valuable resource: it is substantially larger and more text-rich than existing video-centric datasets, it is the first to provide explicit intent annotations at scale, and its retrieval results offer a concrete, reproducible falsifiable claim about a limitation of current video foundation models. The authors also deserve credit for releasing the data, code, and model evaluation results, and the internal statistics (103K samples, 22M words, average 225 words) are consistent across the paper. The central empirical finding—a large retrieval gap between factual and intent queries—is interesting and worth investigating, but its interpretation depends entirely on the validity of the intent labels, which is insufficiently established.

major comments (5)
  1. [§3.2, Step 3 (pre-validation)] The pre-validation step accepts an intent expression only when the embedding similarity between the intent term produced by Qwen2.5-Omni and that of a second, unspecified mLLM exceeds an unreported threshold. This is a measure of inter-model agreement, not of correctness against the true intent of the uploader or protagonist; if the second model shares training data or biases with Qwen2.5-Omni, the agreement is uninformative. The paper must report the second model's identity, the similarity threshold, the extraction procedure, and ideally the agreement of both models with human judgments on a subsample.
  2. [§3.2, Step 3 (post-validation)] The post-validation step generates a 10-second video from each intent expression using Wan2.1 and asks two expert annotators to judge whether the generated video is 'reasonable'. This tests whether the text-to-video model can produce a plausible rendering of the intent expression, not whether the expression reflects the actual purpose of the video creator or the depicted person. The paper provides no criteria for 'reasonable', no inter-annotator agreement, and no account of how failures were treated (discarded, regenerated, or edited). This is a load-bearing weakness because the post-validation is one of only two mechanisms supporting the claim that the intent labels are valid ground truth.
  3. [§3 and §4 (3,000-sample benchmark)] The paper states that all 3,000 benchmark samples were 'independently validated by three professional annotators', but it gives no annotation instructions, no sample selection criteria, and no inter-annotator reliability statistics such as Cohen's kappa or Krippendorff's alpha. Without such evidence, the gold-standard status of the benchmark is unsupported, and the retrieval results in Tables 2 and 3 cannot be interpreted as measuring performance against reliable intent ground truth.
  4. [§3.4 and Table 2] The retrieval comparison confounds the depth of the described cognitive layer with the length and lexical form of the queries. Factual-layer descriptions average 143 words, while abstract- and intent-layer descriptions average 38 and 43 words, respectively (Section 3.4). Longer, more specific queries are generally easier to match to videos, and the fixed intent template '[subject] aims to [intent] by [action]' may be distributionally distant from the natural captions on which the evaluated models were trained. The reported R@1 drop from ~80% to ~35% may therefore reflect text length and format rather than a lack of deep video understanding. The authors should include matched-length query baselines, retrieval using human-written intent descriptions, or a per-layer human upper bound to separate these factors.
  5. [§4 and §5] The paper's broader conclusion that 'current foundation models demonstrate limited capability in representing the latent purposes of videos' (Section 4) is only as strong as the validity of the intent labels. Given that the labels are generated and validated by the methods described in Section 3.2, the 35% R@1 result may partly measure how well models can recover synthetic intent guesses rather than genuine human intent. A experiment in which human annotators retrieve videos from intent queries, or in which intent labels are compared against uploader-provided titles/descriptions, would substantially strengthen the claim.
minor comments (6)
  1. [§1 and Figure 1] The text refers to 'SoRA' in the introduction; this is presumably a typo for 'Sora'.
  2. [§3.1 and Figure 2] The sentence 'The selected quadruples would go through the text generation process (Figure 2)' is repeated almost verbatim in the same paragraph; one occurrence should be removed.
  3. [§3.3] The phrase '6W-element tags' is unclear; consider defining '6W' (who, where, when, why, how, what) explicitly at first use.
  4. [§4] In the sentence 'These data are strictly restricted to be used only for testing and MUST NOT be involved in any stage of model training', the formatting 'MUST NOT' appears to be a typographical error; use normal capitalization.
  5. [References] Reference [2] contains a typo: 'Aannual meeting' should be 'Annual meeting'. Also, many references lack DOIs or arXiv IDs; this is acceptable for a preprint but should be cleaned up for the camera-ready version.
  6. [Table 1] The column header 'Len 𝑡𝑒𝑥𝑡' is awkwardly formatted; use 'Text length' or 'Avg. words'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the retrieval gap is an independent empirical measurement against a separately human-validated 3K benchmark, despite under-specified label-validation methodology.

full rationale

The central claim—that VideoMind supplies deep-cognitive intent annotations and that current video foundation models retrieve intent queries at only ~35% R@1 versus ~80% for factual queries (Table 2)—is an empirical finding, not a quantity derived by construction from the dataset-generation pipeline. Intent labels are generated by Qwen2.5-Omni and filtered by inter-mLLM embedding similarity and by Wan2.1 text-to-video reconstruction (Section 3.2, Step 3), but the 3,000-sample retrieval benchmark is separately and explicitly described as 'independently validated by three professional annotators' (Section 3) and as 'meticulously validated by expert annotators' (Section 4). The evaluated models (InternVideo, UMT-L, CLIP-VIP, mPLUG-2, VAST) are not the label-generating model, and no parameter of those models is fitted to the intent labels. Therefore the R@1 gap is not forced by the paper's own definitions or by a self-citation chain. The validation protocol is under-specified—no embedding-similarity threshold, no annotator instructions, no inter-annotator agreement statistics—which is a genuine correctness and benchmark-validity concern, because the labels may not reflect true uploader or protagonist intent. That concern is about evidential strength, however, not circularity: the retrieval results would remain an independent measurement even if the labels were imperfect. No load-bearing self-citations, imported uniqueness theorems, or ansatz-via-citation steps appear. Accordingly, no circular step can be exhibited, and the appropriate score is 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

The paper introduces no physical or mathematical entities, but it does introduce a new evaluation construct (deep-cognitive understanding) whose ground truth is machine-generated. The free parameters are design choices in the data pipeline, and the axioms are the assumptions that the LLM-generated labels and validation procedures are trustworthy.

free parameters (3)
  • minimum clip duration threshold = 5 seconds
    Hand-chosen cutoff to ensure samples contain sufficient intent information (Section 3.1). Not derived from data.
  • number of sampled key frames = 12
    Randomly select 12 frames per clip as image modality (Section 3.1).
  • intent embedding similarity threshold = not reported
    Pre-validation deems intent correct above an unspecified similarity; threshold is omitted (Section 3.2 Step 3).
assumptions (4)
  • domain assumption mLLM-generated text descriptions are faithful and complete representations of video content across modalities.
    The COT generation pipeline (Qwen2.5-Omni) assumes model outputs are accurate; no human verification for the 100K training samples, only dual-validation for intent on a subset.
  • ad hoc to paper Embedding similarity between two mLLMs' intent speculations is a valid proxy for intent correctness.
    Pre-validation in Section 3.2 Step 3 compares intent terms via embedding similarity; this is internal model agreement, not ground-truth intent.
  • ad hoc to paper Text-to-video generation (Wan2.1) followed by expert judgment can validate intent expressions.
    Post-validation uses generated video reasonableness to assess text quality; this assumes T2V plausibility tracks true intent.
  • domain assumption A 5-second minimum duration guarantees meaningful intent.
    Section 3.1 states clips at least 5 seconds; no evidence for this cutoff.
invented entities (1)
  • deep-cognitive video understanding (as a measurable construct)
    purpose: Provides the evaluative target for the 3K benchmark and frames the dataset's novelty.
    The construct is defined by the paper's own layered annotations; no external operationalization (e.g., correlation with human intent judgments) is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of VideoMind: An Omni-Modal Video Dataset with Intent Grounding for Deep-Cognitive Video Understanding." pith.science (2026). https://pith.science/paper/4ZAVOTUO

@misc{pith2026250718552,
  author       = {Pith},
  title        = {Pith review of: VideoMind: An Omni-Modal Video Dataset with Intent Grounding for Deep-Cognitive Video Understanding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4ZAVOTUO}},
  note         = {Machine review of arXiv:2507.18552}
}
read the original abstract

This paper introduces VideoMind, a video-centric omni-modal dataset designed for deep video content cognition and enhanced multi-modal feature representation. The dataset comprises 103K video samples (3K reserved for testing), each paired with audio and systematically detailed textual descriptions. Specifically, every video and its audio is described across three hierarchical layers (factual, abstract, and intent), progressing from surface to depth. It contains over 22 million words, averaging ~225 words per sample. VideoMind's key distinction from existing datasets is its provision of intent expressions, which require contextual integration across the entire video and are not directly observable. These deep-cognitive expressions are generated using a Chain-of-Thought (COT) approach, prompting the mLLM through step-by-step reasoning. Each description includes annotations for subject, place, time, event, action, and intent, supporting downstream recognition tasks. Crucially, we establish a gold-standard benchmark with 3,000 manually validated samples for evaluating deep-cognitive video understanding. We design hybrid-cognitive retrieval experiments, scored by multi-level retrieval metrics, to appropriately assess deep video comprehension. Evaluation results for models (e.g., InternVideo, VAST, UMT-L) are released. VideoMind serves as a powerful benchmark for fine-grained cross-modal alignment and advances fields requiring in-depth video understanding, such as emotion and intent recognition. The data is publicly available on GitHub, HuggingFace, and OpenDataLab, https://github.com/cdx-cindy/VideoMind.

Figures

Figures reproduced from arXiv: 2507.18552 by the authors.

Figure 1
Figure 1. Examples of video clips in VideoMind (More examples are presented on https://opendatalab.com/Dixin/VideoMind). Abstract This paper introduces VideoMind, a video-centric omni-modal dataset, which enables the deep cognition of video content and enhances feature representations of multi-modal data. The Video￾Mind dataset contains 103K video samples (3K for test only), each of which is accompanied by audio, as well as s… view at source ↗
Figure 2
Figure 2. Text generation pipeline of VideoMind. creators and the primary subjects within the videos. We elaborate in detail on each step of the generation process in the following. Step 1: Multiperspective Descriptions and High-Level Sum￾mary. Unlike previous video-centric datasets, VideoMind hopes to fully reflect information from various sources without any loss. Therefore, it is required that the mLLM separately describe … view at source ↗
Figure 3
Figure 3. Video statistics in VideoMind. Furthermore, we carefully selected 3,000 samples encompassing a broad spectrum of categories. The quality of the intent layer is rigorously evaluated by professional annotators. Subsequently, we established a benchmark serving as the first standardized criterion for the deep understanding of videos. 3.3 Tagging VideoMind further delivers 6W-element tags, including subject (who), place … view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The word cloud of intent, the intent-driven actions, audio style, subject, and place in the VideoMind dataset. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Reading Between the Frames: Interpreting Implicit and Non-literal Meaning in Social Media Videos

    cs.CL 2026-08 conditional novelty 5.0 of 10

    A new video benchmark, DrivelHub+, shows that current video-language models can describe what happens in social media clips but largely fail to infer the implicit humour, irony, or cultural meaning.

Reference graph

Works this paper leans on

27 extracted references · 16 canonical work pages · cited by 1 Pith paper

  1. [1]

    Max Bain, Arsha Nagrani, Gül Varol, and Andrew Zisserman. 2021. Frozen in time: A joint video and image encoder for end-to-end retrieval. InProceedings of the IEEE/CVF international conference on computer vision. 1728–1738

    2021

  2. [2]

    David Chen and William B Dolan. 2011. Collecting highly parallel data for paraphrase evaluation. InAannual meeting of the association for computational linguistics: human language technologies. 190–200

    work page 2011

  3. [3]

    Sihan Chen, Handong Li, Qunbo Wang, Zijia Zhao, Mingzhen Sun, Xinxin Zhu, and Jing Liu. 2024. Vast: A vision-audio-subtitle-text omni-modality foundation model and dataset.Advances in Neural Information Processing Systems36 (2024)

    work page 2024

  4. [4]

    Chaoyou Fu, Yuhan Dai, Yongdong Luo, Lei Li, Shuhuai Ren, Renrui Zhang, Zihan Wang, Chenyu Zhou, Yunhang Shen, Mengdan Zhang, et al. 2025. Video-mme: The first-ever comprehensive evaluation benchmark of multi-modal llms in video analysis. InProceedings of the Computer Vision and Pattern Recognition Conference. 24108–24118

    work page 2025

  5. [5]

    Xiaowei Hu, Zhe Gan, Jianfeng Wang, Zhengyuan Yang, Zicheng Liu, Yumao Lu, and Lijuan Wang. 2022. Scaling up vision-language pre-training for image captioning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 17980–17989

    work page 2022

  6. [6]

    Ranjay Krishna, Kenji Hata, Frederic Ren, Li Fei-Fei, and Juan Carlos Niebles

  7. [7]

    Jiapeng Li, Ping Wei, Wenjuan Han, and Lifeng Fan. 2023. Intentqa: Context-aware video intent reasoning. InProceedings of the IEEE/CVF international conference on computer vision. 11963–11974

    work page 2023

  8. [8]

    Kunchang Li, Yali Wang, Yizhuo Li, Yi Wang, Yinan He, Limin Wang, and Yu Qiao

    work page

Show all 27 references
  1. [9]

    Bin Lin, Yang Ye, Bin Zhu, Jiaxi Cui, Munan Ning, Peng Jin, and Li Yuan. 2024. Video-LLaVA: Learning United Visual Representation by Alignment Before Pro- jection. InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 5971–5984

    2024

  2. [10]

    Jing Liu, Sihan Chen, Xingjian He, Longteng Guo, Xinxin Zhu, Weining Wang, and Jinhui Tang. 2024. Valor: Vision-audio-language omni-perception pretraining model and dataset.IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    2024

  3. [11]

    Antoine Miech, Dimitri Zhukov, Jean-Baptiste Alayrac, Makarand Tapaswi, Ivan Laptev, and Josef Sivic. 2019. Howto100m: Learning a text-video embedding by watching hundred million narrated video clips. InProceedings of the IEEE/CVF international conference on computer vision. 2630–2640

    2019

  4. [12]

    Kepan Nan, Rui Xie, Penghao Zhou, Tiehan Fan, Zhenheng Yang, Zhijie Chen, Xiang Li, Jian Yang, and Ying Tai. 2024. Openvid-1m: A large-scale high-quality dataset for text-to-video generation.arXiv preprint arXiv:2407.02371(2024)

    2024 arXiv

  5. [13]

    Anna Rohrbach, Atousa Torabi, Marcus Rohrbach, Niket Tandon, Christopher Pal, Hugo Larochelle, Aaron Courville, and Bernt Schiele. 2017. Movie description. International Journal of Computer Vision123 (2017), 94–120

    2017

  6. [14]

    Ramon Sanabria, Ozan Caglayan, Shruti Palaskar, Desmond Elliott, Lo"ic Bar- rault, Lucia Specia, and Florian Metze. 2018. How2: A Large-scale Dataset For Multimodal Language Understanding. InProceedings of the Workshop on Visually Grounded Interaction and Language (ViGIL). Neu...

    2018

  7. [15]

    Bart Thomee, David A Shamma, Gerald Friedland, Benjamin Elizalde, Karl Ni, Douglas Poland, Damian Borth, and Li-Jia Li. 2016. Yfcc100m: The new data in multimedia research.Commun. ACM59, 2 (2016), 64–73

    2016

  8. [16]

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, et al. 2024. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv preprint arXiv:2409.12191(2024)

    2024 arXiv

  9. [17]

    Xin Wang, Jiawei Wu, Junkun Chen, Lei Li, Yuan-Fang Wang, and William Yang Wang. 2019. VaTeX: A Large-Scale, High-Quality Multilingual Dataset for Video- and-Language Research. InThe IEEE International Conference on Computer Vision (ICCV)

    2019

  10. [18]

    Yi Wang, Yinan He, Yizhuo Li, Kunchang Li, Jiashuo Yu, Xin Ma, Xinyuan Chen, Yaohui Wang, Ping Luo, Ziwei Liu, Yali Wang, Limin Wang, and Yu Qiao. 2023. InternVid: A Large-scale Video-Text Dataset for Multimodal Understanding and Generation.arXiv preprint arXiv:2307.06942(2023...

    2023 arXiv

  11. [19]

    Yi Wang, Kunchang Li, Yizhuo Li, et al . 2022. Internvideo: General video foundation models via generative and discriminative learning.arXiv preprint arXiv:2212.03191(2022)

    2022 arXiv

  12. [20]

    Haiyang Xu, Qinghao Ye, Ming Yan, Yaya Shi, Jiabo Ye, Yuanhong Xu, Chenliang Li, Bin Bi, et al. 2023. mplug-2: A modularized multi-modal foundation model across text, image and video. InInternational Conference on Machine Learning. 38728–38748

    2023

  13. [21]

    Jun Xu, Tao Mei, Ting Yao, and Yong Rui. 2016. Msr-vtt: A large video description dataset for bridging video and language. InIEEE/CVF Conference on Computer Vision and Pattern Recognition. 5288–5296

    2016

  14. [22]

    Hongwei Xue, Tiankai Hang, Yanhong Zeng, Yuchong Sun, Bei Liu, Huan Yang, Jianlong Fu, and Baining Guo. 2022. Advancing high-resolution video- language representation with large-scale video transcriptions. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern...

    2022

  15. [23]

    Hongwei Xue, Yuchong Sun, Bei Liu, Jianlong Fu, Ruihua Song, Houqiang Li, and Jiebo Luo. 2023. Clip-vip: Adapting pre-trained image-text model to video- language representation alignment. InInternational Conference on Learning Rep- resentation

    2023

  16. [24]

    Rowan Zellers, Ximing Lu, Jack Hessel, Youngjae Yu, Jae Sung Park, Jize Cao, Ali Farhadi, and Yejin Choi. 2021. Merlot: Multimodal neural script knowledge models.Advances in neural information processing systems34 (2021), 23634– 23651

    2021

  17. [25]

    Xin Zhang, Yanzhao Zhang, Wen Xie, Mingxin Li, Ziqi Dai, Dingkun Long, Pengjun Xie, Meishan Zhang, Wenjie Li, and Min Zhang. 2024. GME: Im- proving Universal Multimodal Retrieval by Multimodal LLMs.arXiv preprint arXiv:2412.16855(2024)

    2024 arXiv

  18. [2017]

    InIEEE International Conference on Computer Vision

    Dense-captioning events in videos. InIEEE International Conference on Computer Vision. 706–715

  19. [2023]

    In IEEE/CVF International Conference on Computer Vision

    Unmasked teacher: Towards training-efficient video foundation models. In IEEE/CVF International Conference on Computer Vision. 19948–19960

Pith tools

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