Pith. sign in

REVIEW 3 major objections 4 minor 38 references

T2VParser: Adaptive Decomposition Tokens for Partial Alignment in Text to Video Retrieval

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read Decomposing text and video into shared semantic views lets retrieval align only the parts that match.

desk verdict The core idea is sound, but the headline SOTA numbers rest on a backbone mismatch that needs fixing before the paper's empirical claims can be trusted. read the letter →

arxiv 2507.20518 v1 pith:TWHNS5LC submitted 2025-07-28 cs.CV cs.MM

classification cs.CVcs.MM
keywords text-to-videoretrievalpartialalignmentmultiviewembeddingsAdaptiveDecompositionTokensDualCommunicationMechanismCLIPknowledgetransfervideo-textpretrainingrepresentationdiversity
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 standard text-to-video retrieval trains on wrong supervision because a caption usually describes only part of a video, yet the whole text embedding is pushed toward the whole video embedding. It proposes T2VParser, which uses a small set of learnable tokens shared by both modalities to decompose text and video into multiview embeddings and then aligns only the corresponding views. A Dual Communication Mechanism weights the views before matching, and a diversity loss keeps the views from collapsing into a single representation. The paper reports consistent top-1 recall gains over its own baselines on MSR-VTT, MSVD, DiDeMo, and ActivityNet, with larger gains on longer texts and on noisy or partially relevant queries.

What carries the argument

The load-bearing object is the Adaptive Decomposition Token (ADT): a set of $k$ learnable tokens shared across modalities, used as queries in DETR-style cross-attention parsers that pull distinct semantic views out of the video and text representations. The same token index on the two sides is meant to correspond to the same view, which is what makes comparing the resulting embeddings meaningful. The Dual Communication Mechanism then exchanges information between the two modalities, extracts the diagonal of a video–text similarity matrix as a soft weighting over views, and produces the aggregated representations $\tilde f_v$ and $\tilde f_t$ used for retrieval. A representation diversity loss keeps the $k$ views mutually dissimilar, preventing the parser from collapsing all queries into one generic embedding.

What would settle it

If, after permuting the order of the decomposition tokens on only the text side (or only the video side) at inference, retrieval recall is unchanged, then same-index correspondence between views carries no signal and the performance would be attributable to the attention-weighted aggregation rather than to the claimed semantic decomposition.

Watch

Extended reading notes

Core claim

The central claim is that partial alignment—matching only the semantically overlapping components of a video and its caption—is both necessary and sufficient for better text-to-video retrieval, and that a shared set of Adaptive Decomposition Tokens can produce those components. On each side, the same $k$ learnable query tokens attend over the pretrained encoder's video or text features, yielding $k$ perspective-specific embeddings that are concatenated with the encoder's local features. The Dual Communication Mechanism computes a similarity matrix between the two sets of views, turns it into attention weights, and aggregates each side so irrelevant views are down-weighted instead of being forced to match. Contrastive training on the aggregated representations, plus a diversity loss on the views, gives the final retrieval score at inference. In the paper's experiments the gains grow as the text becomes richer, from roughly one recall point on the short-caption MSR-VTT-1k setting to over three points on ActivityNet, and the same model attached to CLIP4Clip, CLIP-VIP, or Mug-STAN improves each baseline.

Load-bearing premise

The method assumes that the same learnable token, after separately attending to text and to video, comes to represent the same semantic view in both modalities, so that comparing same-index views is meaningful.

Editorial extensions

If this is right

  • T2VParser can be attached to existing CLIP-based video–text encoders without retraining the pretrained backbone, and it improves top-1 recall for each of the four encoder/baseline pairs tested.
  • The benefit is largest when the input content is information-rich: gains on DiDeMo, ActivityNet, and long-text versions of MSR-VTT and MSVD exceed gains on the short-caption MSR-VTT-1k split.
  • On partially relevant data, such as ActivityNet queries with 25 percent of captions replaced by captions from other videos and the PRVR benchmark, T2VParser reduces the retrieval drop relative to the same baselines.
  • Inference still uses a single similarity score between the two aggregated view representations, so the framework changes how representations are built rather than how retrieval is performed.
  • Without the representation diversity loss, the decomposition tokens collapse toward redundant embeddings and recall drops, indicating that view separation is needed for the alignment mechanism to help.

Reading between the lines

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

  • A direct test of the token-correspondence assumption is to permute the ADT order on only one modality at inference: if same-index views are truly aligned, recall should drop noticeably under permutation; if not, the gains likely come from attention-weighted pooling rather than semantic decomposition.
  • The same shared-token decomposition idea could transfer to other retrieval settings where one side contains strictly more content than the other, such as image–text retrieval with long captions, document–query retrieval, or moment localization inside long videos.
  • Because no explicit cross-modal correspondence loss is applied to the tokens, the paper's framing is stronger than what the training objective enforces; probing the learned views with caption fragments or frame subsets would show which views actually specialize.
  • The document-generation step depends on an LLM's summary style, and the paper's ablation shows LLM choice matters little; an interesting extension is whether the structural prompt constraints, rather than the LLM itself, are what the parser benefits from.
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

3 major / 4 minor

Summary. The paper proposes T2VParser, a model-agnostic framework for text-to-video retrieval that decomposes video and text into multiview embeddings using a set of modality-shared learnable Adaptive Decomposition Tokens (ADTs). A Dual Communication Mechanism exchanges and filters information between the resulting view sets, and training combines contrastive alignment with a diversity loss and with synthetic 'video documents' generated by key-frame captioning and LLM summarization. The authors report consistent gains over several baselines on MSR-VTT, MSVD, DiDeMo, and ActivityNet, as well as on self-created long-text versions of MSR-VTT and MSVD, and they claim state-of-the-art performance.

Significance. If the reported gains hold under matched experimental conditions, the ADT decomposition and Dual Communication Mechanism constitute a reusable approach to partial text-video alignment that preserves the knowledge of pretrained image-text encoders. The paper includes ablations for the main components, releases code, and evaluates robustness to noisy and partially relevant content, which is a practically relevant scenario. However, the headline state-of-the-art claim currently rests on comparisons that mix backbones and the DSL trick, and the central premise that same-index token views correspond semantically across modalities is not validated. The experiments also lack error bars or significance testing, which matters because several controlled gains are only about one R@1 point.

major comments (3)
  1. [Section 4.2, Tables 1 and 2] The claimed state-of-the-art and consistent improvement results are not established under matched experimental conditions. In Table 1, T2VParser+Mug-STAN and T2VParser+CLIP-VIP report 58.4 R@1 on MSR-VTT-1k against baselines of 57.3 and 57.7, but those T2VParser numbers appear in Table 2 only in the CLIP-ViT-B/16 rows with the DSL trick (T2VParser+Mug-STAN†* and T2VParser+CLIP-VIP†*), while the Table 1 baselines are not identified as B/16 or B/32 or as with/without DSL. In the controlled B/32 rows of Table 2, T2VParser+Mug-STAN† improves only from 50.9 to 51.6 and T2VParser+CLIP-VIP† from 50.1 to 51.3, which are far below published baseline numbers. The absolute state-of-the-art claim is also contradicted by ActivityNet, where CLIP-VIP* (61.4) exceeds T2VParser+Mug-STAN†* (60.5). Please report every comparison with matched backbone, input resolution, and DSL usage, and avoid claiming state-of-the-art on the basis of the mixed-condition numbers in Table 1.
  2. [Section 3.1, Eqs. (1)-(5) and Section 3.2, Eqs. (6)-(10)] The method assumes that the i-th adaptive decomposition token in the video parser and the i-th token in the text parser capture corresponding semantic views, because the final matching score uses the diagonal of the similarity matrix in Eq. (9). Nothing in the formulation enforces or measures this cross-modal correspondence: the diversity loss in Eq. (17) separates views only within each modality, and the parsers are trained independently with supervision applied only to the final aggregated vectors e_fv and e_ft. Figure 4 demonstrates within-modality diversity but does not show cross-modal view correspondence. This is a load-bearing assumption for the paper's 'partial alignment' interpretation. Please provide quantitative evidence of cross-modal token correspondence (for example, per-view retrieval analysis, attention-overlap statistics, or canonical correlation), or reformulate the claim so that it does not depend on same-index semantic matching.
  3. [Section 4.1 and Section 4.2] The central empirical demonstration relies on self-created long-text datasets (MSR-VTT-Doc and MSVD-Concat), but these datasets are not released and the code repository link does not guarantee access to them. Since the paper argues that the method's advantage grows with text complexity, the long-text benchmarks are necessary for reproduction. In addition, all reported numbers appear to be single runs with no error bars or significance testing; given that the matched-condition gains in Table 2 are around 0.7 to 1.9 R@1 points, the paper should report multiple seeds and variance, or a significance test. Please release the generated long-text datasets (or the full generation pipeline with seeds) and add variance estimates.
minor comments (4)
  1. [Section 5 and Table 7] The text in Section 5 says T2VParser introduces a 15% longer per-query retrieval time, but Table 7 shows 283 ms versus 192 ms for CLIP-VIP (47% longer) and 834 ms versus 584 ms (43% longer). Please reconcile the text with the table.
  2. [References] References [21] and [22] are duplicates: both entries list the same title, authors, venue, and page range for 'Revisiting Temporal Modeling for CLIP-Based Image-to-Video Knowledge Transferring', yet Table 2 cites [22] as X-CLIP (CVPR'23). The X-CLIP citation should point to the correct paper.
  3. [Implementation Details and Section 4.1] There is a discrepancy about when the generated documents are used: Section 4 says 'the extra captions only use in training stage', but Section 4.1 constructs MSR-VTT-Doc by generating documents for both training and testing splits and evaluates on them. Please clarify whether the long-text test sets are used at test time and how this is consistent with the statement that extra captions are training-only.
  4. [Tables] Several table captions and headers contain fragments rather than complete sentences, such as Table 1's caption ending with 'MSVD-Concat(288.3)' and Table 2's stray 'Caption' column label. Please clean up the table formatting.

Circularity Check

0 steps flagged · score 1.0 of 10

No material circularity: T2VParser is an empirical contrastive retrieval pipeline whose headline results are benchmark measurements, with only a non-load-bearing self-citation.

full rationale

T2VParser makes no first-principles derivation claim; its central assertions are retrieval performance numbers on MSR-VTT, MSVD, DiDeMo, and ActivityNet, measured against published baselines under standard evaluation protocols. The shared Adaptive Decomposition Tokens and the Dual Communication Mechanism are architectural proposals, and the fact that token correspondence across modalities is not explicitly enforced is an untested assumption rather than a circular dependency. The contrastive objective in Eqs. 11-13 optimizes the same embeddings that are later ranked, which is the normal self-supervised/contrastive training paradigm, not a prediction reducing to its input. The one self-citation, [25], is used to motivate the representation diversity loss: the paper states that 'using only alignment loss may cause feature collapse [25]'. This citation is not load-bearing because the paper supplies its own supporting evidence in Figure 4 and in the Table 5 ablation 'w/o Representation Diversity Loss', and the central SOTA claim rests on independent benchmark evaluations rather than on this citation. The experimental-comparison critique about mismatched backbones and DSL tricks in Table 2 is a correctness and experimental-design concern, not a circularity reduction. Overall, the derivation chain is self-contained in the sense relevant to circularity: no fitted parameter is renamed as a prediction, no uniqueness theorem is imported from the authors' prior work, and no known result is repackaged as a derivation.

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

The central claim relies on several hyperparameters (number of tokens, diversity weight, parser depth, key-frame count) and on assumptions about pretrained feature sufficiency, cross-modal token correspondence, and faithfulness of generated documents. The cross-modal correspondence assumption is the most fragile because nothing in the loss explicitly aligns token indices across modalities.

free parameters (5)
  • Number of Adaptive Decomposition Tokens k = 8
    Chosen in Section 4.7 as the peak recall point on MSR-VTT 1k; central claim depends on this capacity choice.
  • Diversity loss weight alpha = 0.1
    Set in Section 4 implementation details; no sensitivity study is reported.
  • Parser depth = 8 layers
    Selected after parameter analysis in Section 4.7; deeper configurations give the best reported performance.
  • Key frames per video = 8
    Used for document generation in Section 3.3; no ablation is reported on this count.
  • Text token length and frame sampling = 128 tokens, 12 frames (64 for long videos)
    Implementation choices in Section 4; standard values for this literature but still chosen rather than derived.
assumptions (4)
  • domain assumption CLIP/STAN pretrained features contain enough semantic information for ADTs to extract separable views.
    Section 3.1 selects STAN as the feature extractor and relies on its representations as the source content for cross-attention; if the base features lack the semantics, the parser cannot recover them.
  • domain assumption A shared token index corresponds to the same semantic perspective across modalities.
    Section 3.1 and Figure 1(d) assume same-token outputs align cross-modally, but no constraint enforces this correspondence; this is the weakest assumption.
  • domain assumption Text captions are only partially aligned with video content, and partial alignment is the main source of retrieval error.
    Section 1 uses similarity histograms to argue information inequivalence; the whole method is designed around this premise.
  • domain assumption Generated video documents are faithful enough to serve as additional positive queries without introducing harmful noise.
    Section 3.3 builds documents from key frames, BLIP-2 captions, and an LLM; ablations in Table 5 partially support the benefit but do not characterize failure cases.

how reviews work

0 comments
Cite this review

Pith. "Pith review of T2VParser: Adaptive Decomposition Tokens for Partial Alignment in Text to Video Retrieval." pith.science (2026). https://pith.science/paper/TWHNS5LC

@misc{pith2026250720518,
  author       = {Pith},
  title        = {Pith review of: T2VParser: Adaptive Decomposition Tokens for Partial Alignment in Text to Video Retrieval},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TWHNS5LC}},
  note         = {Machine review of arXiv:2507.20518}
}
read the original abstract

Text-to-video retrieval essentially aims to train models to align visual content with textual descriptions accurately. Due to the impressive general multimodal knowledge demonstrated by image-text pretrained models such as CLIP, existing work has primarily focused on extending CLIP knowledge for video-text tasks. However, videos typically contain richer information than images. In current video-text datasets, textual descriptions can only reflect a portion of the video content, leading to partial misalignment in video-text matching. Therefore, directly aligning text representations with video representations can result in incorrect supervision, ignoring the inequivalence of information. In this work, we propose T2VParser to extract multiview semantic representations from text and video, achieving adaptive semantic alignment rather than aligning the entire representation. To extract corresponding representations from different modalities, we introduce Adaptive Decomposition Tokens, which consist of a set of learnable tokens shared across modalities. The goal of T2VParser is to emphasize precise alignment between text and video while retaining the knowledge of pretrained models. Experimental results demonstrate that T2VParser achieves accurate partial alignment through effective cross-modal content decomposition. The code is available at https://github.com/Lilidamowang/T2VParser.

Figures

Figures reproduced from arXiv: 2507.20518 by the authors.

Figure 1
Figure 1. (a) Text length distributions; (b) Frame-text similar [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. An overview of our T2VParser. (a) illustrates the processing and training pipeline of T2VParser. It shows how the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The pipeline of video document generation. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Impact of Modality Content on Multiview Embed [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Different Number of Learnable Queries and Layers [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 32 canonical work pages

  1. [1]

    Lisa Anne Hendricks, Oliver Wang, Eli Shechtman, Josef Sivic, Trevor Darrell, and Bryan Russell. 2017. Localizing moments in video with natural language. In 2017 IEEE international conference on computer vision . 5803–5812

  2. [2]

    Max Bain, Arsha Nagrani, Gül Varol, and Andrew Zisserman. 2021. Frozen in Time: A Joint Video and Image Encoder for End-to-End Retrieval. In IEEE/CVF International Conference on Computer Vision, 2021 . 1708–1718

  3. [3]

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexan- der Kirillov, and Sergey Zagoruyko. 2020. End-to-End Object Detection with Transformers. In Computer Vision - 16th European Conference, 2020 , Vol. 12346. 213–229

  4. [4]

    David Chen and William B Dolan. 2011. Collecting highly parallel data for paraphrase evaluation. In Proceedings of the 49th annual meeting of the association for computational linguistics: human language technologies . 190–200

  5. [5]

    Shizhe Chen, Yida Zhao, Qin Jin, and Qi Wu. 2020. Fine-Grained Video-Text Retrieval With Hierarchical Graph Reasoning. In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020 . 10635–10644

  6. [6]

    Xing Cheng, Hezheng Lin, Xiangyu Wu, Fan Yang, and Dong Shen. 2021. Improv- ing Video-Text Retrieval by Multi-Stream Corpus Alignment and Dual Softmax Loss. arXiv preprint arXiv:2109.04290 (2021)

  7. [7]

    DeepSeek-AI, Aixin Liu, Bei Feng, Bin Wang, Bingxuan Wang, Bo Liu, Chenggang Zhao, Chengqi Deng, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, Deli Chen, Dongjie Ji, Erhang Li, Fangyun Lin, Fuli Luo, Guangbo Hao, Guanting Chen, Guowei Li, Hao Zhang, Hanwei Xu, Hao Yang, Haowei Zhang, Honghui Ding, Huajian Xin, Huazuo Gao, Hui Li, Hui Qu, J. L. Cai, Jian L...

  8. [8]

    Jianfeng Dong, Xianke Chen, Minsong Zhang, Xun Yang, Shujie Chen, Xirong Li, and Xun Wang. 2022. Partially Relevant Video Retrieval. In MM ’22: The 30th ACM International Conference on Multimedia . 246–257

Show all 38 references
  1. [9]

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, and et al. 2024. The Llama 3 Herd of Models. arXiv preprint arXiv:2407.21783 (2024)

  2. [10]

    Valentin Gabeur, Chen Sun, Karteek Alahari, and Cordelia Schmid. 2020. Multi- modal Transformer for Video Retrieval. In Computer Vision - ECCV 2020 - 16th European Conference. 214–229

  3. [11]

    Team GLM, Aohan Zeng, Bin Xu, and et al. 2024. ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools. arXiv:2406.12793

  4. [12]

    Satya Krishna Gorti, Noël Vouitsis, Junwei Ma, Keyvan Golestan, Maksims Volkovs, Animesh Garg, and Guangwei Yu. 2022. X-Pool: Cross-Modal Language- Video Attention for Text-Video Retrieval. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022 . 4996–5005

  5. [13]

    Chan Hur, Jeong hun Hong, Dong hun Lee, Dabin Kang, Semin Myeong, Sang hyo Park, and Hyeyoung Park. 2025. Narrating the Video: Boosting Text-Video Retrieval via Comprehensive Utilization of Frame-Level Captions. Proceedings of the IEEE/CVF Conference on Computer Vision and Pat...

  6. [14]

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

  7. [15]

    Jie Lei, Linjie Li, Luowei Zhou, Zhe Gan, Tamara L Berg, Mohit Bansal, and Jingjing Liu. 2021. Less is more: Clipbert for video-and-language learning via sparse sampling. In IEEE conference on computer vision and pattern recognition . 7331–7341

  8. [16]

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven C. H. Hoi. 2023. BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models. In International Conference on Machine Learning, 2023 (Proceedings of Machine Learning Research, Vol. 202)...

  9. [17]

    Tianhong Li, Sangnie Bhardwaj, Yonglong Tian, Han Zhang, Jarred Barber, Dina Katabi, Guillaume Lajoie, Huiwen Chang, and Dilip Krishnan. 2024. Leveraging Unpaired Data for Vision-Language Generative Models via Cycle Consistency. In The Twelfth International Conference on Learn...

  10. [18]

    Yili Li, Jing Yu, Keke Gai, Bang Liu, Gang Xiong, and Qi Wu. 2024. T2VIndexer: A Generative Video Indexer for Efficient Text-Video Retrieval. In Proceedings of the 32nd ACM International Conference on Multimedia, MM, 2024 . 3955–3963

  11. [19]

    Chengzhi Lin, Ancong Wu, Junwei Liang, Jun Zhang, Wenhang Ge, Wei-Shi Zheng, and Chunhua Shen. 2022. Text-Adaptive Multiple Visual Prototype Matching for Video-Text Retrieval. In Advances in Neural Information Processing Systems 35: Annual Conference on Neural Information Proc...

  12. [20]

    Ruyang Liu, Jingjia Huang, Wei Gao, Thomas H Li, and Ge Li. 2023. Mug-STAN: Adapting Image-Language Pretrained Models for General Video Understanding. arXiv preprint arXiv:2311.15075 (2023)

  13. [22]

    Ruyang Liu, Jingjia Huang, Ge Li, Jiashi Feng, Xinglong Wu, and Thomas H. Li

  14. [23]

    Yang Liu, Samuel Albanie, Arsha Nagrani, and Andrew Zisserman. 2019. Use What You Have: Video retrieval using representations from collaborative experts. In 30th British Machine Vision Conference 2019 . 279

  15. [24]

    Huaishao Luo, Lei Ji, Ming Zhong, Yang Chen, Wen Lei, Nan Duan, and Tianrui Li. 2022. CLIP4Clip: An empirical study of CLIP for end to end video clip retrieval and captioning. Neurocomputing 508 (2022), 293–304

  16. [25]

    In IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Revisiting Temporal Modeling for CLIP-Based Image-to-Video Knowledge Transferring. In IEEE/CVF Conference on Computer Vision and Pattern Recognition . 6555–6564

  17. [26]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. Learning Transferable Visual Models From Natural Language Supervision. In Proceedings...

  18. [27]

    de Melo, Benjamin Van Durme, and Rama Chellappa

    Arun Reddy, Alexander Martin, Eugene Yang, Andrew Yates, Kate Sanders, Ken- ton Murray, Reno Kriz, Celso M. de Melo, Benjamin Van Durme, and Rama Chellappa. 2025. Video-ColBERT: Contextualized Late Interaction for Text-to- Video Retrieval. Proceedings of the IEEE/CVF Conferenc...

  19. [28]

    Xiangyan Qu, Jing Yu, Keke Gai, Jiamin Zhuang, Yuanmin Tang, Gang Xiong, Gaopeng Gou, and Qi Wu. 2024. Visual-Semantic Decomposition and Partial Alignment for Document-based Zero-Shot Learning. In Proceedings of the 32nd ACM International Conference on Multimedia, 2024 . 4581–4590

  20. [29]

    Zeyu Wang, Yu Wu, Karthik Narasimhan, and Olga Russakovsky. 2022. Multi- query Video Retrieval. In Computer Vision-17th European Conference, 2022 , Vol. 13674. 233–249

  21. [30]

    Peng Wu, Xiangteng He, Mingqian Tang, Yiliang Lv, and Jing Liu. 2021. HANet: Hierarchical Alignment Networks for Video-Text Retrieval. In ACM Multimedia Conference, 2021. 3518–3527

  22. [31]

    Hao Tang, Lei Ding, Songsong Wu, Bin Ren, Nicu Sebe, and Paolo Rota. 2023. Deep Unsupervised Key Frame Extraction for Efficient Video Classification.ACM Trans. Multim. Comput. Commun. Appl. 19, 3 (2023), 119:1–119:17

  23. [32]

    Jun Xu, Tao Mei, Ting Yao, and Yong Rui. 2016. Msr-vtt: A large video description dataset for bridging video and language. In 2016 IEEE conference on computer vision and pattern recognition . 5288–5296

  24. [33]

    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 Alignment. In The Eleventh International Conference on Learning Repre- sentations, 2023

  25. [34]

    Wenhao Wu, Haipeng Luo, Bo Fang, Jingdong Wang, and Wanli Ouyang. 2023. Cap4Video: What Can Auxiliary Captions Do for Text-Video Retrieval?. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

  26. [35]

    Haonan Zhang, Pengpeng Zeng, Lianli Gao, Jingkuan Song, and Heng Tao Shen

  27. [37]

    Bowen Zhang, Hexiang Hu, and Fei Sha. 2018. Cross-modal and hierarchical modeling of video and text. In 2018 european conference on computer vision . 374– 390

  28. [2017]

    In IEEE International Conference on Computer Vision 2017

    Dense-Captioning Events in Videos. In IEEE International Conference on Computer Vision 2017. 706–715

  29. [2023]

    In IEEE/CVF Conference on Computer Vision and Pattern Recognition,

    Revisiting Temporal Modeling for CLIP-Based Image-to-Video Knowledge Transferring. In IEEE/CVF Conference on Computer Vision and Pattern Recognition,

  30. [2024]

    InProceedings of the 32nd ACM International Conference on Multimedia

    MPT: Multi-grained Prompt Tuning for Text-Video Retrieval. InProceedings of the 32nd ACM International Conference on Multimedia . 1206–1214

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.