Pith. sign in

REVIEW 5 major objections 5 minor 49 references

MGCA-Net: Multi-Grained Category-Aware Network for Open-Vocabulary Temporal Action Localization

T0 review · 5 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read MGCA-Net decouples base and novel action recognition by granularity, reporting that supervising base actions at snippet level and novel actions via a coarse-to-fine video-to-proposal classifier raises novel-category localization by more tha

desk verdict Solid OV-TAL/ZS-TAL architecture paper with genuine but modest gains; the transfer assumption in the novel-action path is the main soft spot. read the letter →

arxiv 2511.13039 v2 pith:Q36APAI3 submitted 2025-11-17 cs.CV

Zhenying Fang , Richang Hong This is my paper · ORCID
classification cs.CV
keywords open-vocabularytemporalactionlocalizationzero-shotmulti-grainedcategoryawarenesscoarse-to-fineclassificationpresencepredictionvision-languagemodelscontrastivelearningvideounderstanding
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

Open-vocabulary temporal action localization asks a model to find and label actions in untrimmed video, including categories it was never trained on. Existing methods classify proposals at a single granularity, which the authors argue hurts both familiar and unfamiliar actions. MGCA-Net splits the problem: a supervised classifier handles base (seen) actions at snippet granularity, while a coarse-to-fine classifier handles novel (unseen) actions by first deciding which categories occur anywhere in the video and then assigning each proposal one of those categories. The paper reports that this split improves both base and novel action localization, with novel-category gains of over six points on THUMOS'14 and over three points on ActivityNet-1.3. If the results hold, the insight is that decoupling category perception by granularity makes better use of a vision-language model's zero-shot ability than classifying every proposal against the full novel vocabulary at once.

What carries the argument

The load-bearing component is the coarse-to-fine classifier for novel actions, gated by the action-presence predictor. For each video, the classifier computes image-text similarities from a frozen vision-language model, aggregates them by multi-instance learning over the top eighth of frames, and keeps the top-N categories as coarse categories. Each novel proposal's feature is the mean of frame image features inside its temporal interval, passed through a two-layer projection layer, and the proposal is assigned the coarse category with the highest similarity. The projection layer is the only learned part of the novel path; it is trained with a contrastive loss whose positive text feature is

What would settle it

Measure the recall of the coarse-category step: for videos containing novel actions, check how often the true category appears among the top-N coarse categories selected by the multi-instance learning step. If this recall is far below the reported novel accuracy, or if replacing the learned projection layer with raw vision-language similarity produces the same novel accuracy, then the coarse-to-fine classifier is not the source of the reported gains.

Watch

Extended reading notes

Core claim

The central claim is that multi-grained category awareness is the mechanism that makes open-vocabulary localization work. MGCA-Net uses a category-agnostic localizer to propose candidate segments. An action-presence score and a conventional classifier then divide proposals into base and novel groups: proposals with high action presence and a confident base label become base instances; proposals with high action presence but no confident base label become novel proposals. The novel path operates in two steps. At video granularity, a multi-instance learning step over per-frame image-text similarities selects the coarse categories present in the video. At proposal granularity, averaged image fe

Load-bearing premise

The novel-action path assumes that the contrastive alignment learned on base categories transfers to unseen categories, and that the localizer, trained with unannotated novel instances treated as negatives, still recalls novel action segments at all.

Editorial extensions

If this is right

  • A model can recognize actions it was never annotated for, as long as the action category can be described in text and a frozen vision-language model can relate that text to video frames.
  • Base and novel actions no longer compete in the same classifier, reducing the tendency to misclassify novel proposals as base actions and improving base-category accuracy at the same time.
  • The novel path is mostly training-free, so adapting to a new set of novel categories at inference requires only new text descriptions, not re-training.
  • The same architecture transfers to zero-shot temporal action localization by removing the conventional classifier and action-presence predictor, with the largest gains at high temporal intersection-over-union thresholds.
  • The design suggests that granularity, not just feature quality, is a first-order choice for open-vocabulary video tasks.

Reading between the lines

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

  • My inference: the coarse stage sets a hard upper bound on novel recall; if a true novel category is missing from the top-N coarse categories, the fine stage cannot recover it. A direct test is to measure coarse-category recall against an oracle that always includes the true categories.
  • My inference: because only the projection layer is trained, the method points to a cheap adaptation recipe—keep the vision-language model frozen and train only the gating and projection heads. The same recipe could be tried on other video grounding tasks with minimal retraining.
  • My inference: the paper's stated limitation that unannotated action instances are treated as negatives during localizer training means the reported gap may depend on annotation completeness; on datasets with denser novel-action annotations, the advantage of the two-stage design could narrow or widen.
  • My inference: the improvement at high tIoU thresholds suggests the two-stage assignment mostly reduces category-label noise on novel proposals; a finer error analysis separating temporal misalignment from category confusion would test this directly.
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 / 5 minor

Summary. MGCA-Net addresses open-vocabulary temporal action localization by combining a category-agnostic ActionFormer localizer with an action presence predictor, a conventional base-action classifier, and a coarse-to-fine classifier that uses frozen CLIP image/text features. Novel categories are handled by computing video-level coarse categories via MIL over image-text similarities and then assigning each proposal to one of the coarse categories using a learned projection layer trained with contrastive loss on base categories. The paper reports state-of-the-art OV-TAL and ZS-TAL results on THUMOS'14 and ActivityNet-1.3 under 75/25 and 50/50 category splits, with ablations of each module.

Significance. If the empirical claims hold, the multi-granularity decoupling of base and novel action prediction and the training-free coarse category stage are a useful contribution to OV-TAL. The paper evaluates both OV-TAL and ZS-TAL, does not require prompt tuning, and releases code. However, the evidence is weakened by missing variance information, hyperparameter selection on the benchmark itself, and an untested base-to-novel transfer assumption in the projection layer, so the strength of the claimed state-of-the-art advantage is not yet fully established.

major comments (5)
  1. [Sec. IV-A and Tables I-II] The paper states 'we perform 10 random splits to make it statistically robust' but reports only averaged mAP without standard deviations or per-split results. Several SOTA margins are small: Table II, 75%-25% ActivityNet average mAP is 35.0 for MGCA-Net versus 34.7 for Ti-FAD, and the 50%-50% comparison is 32.2 versus 32.0. Without a measure of spread or a significance test, the 'state-of-the-art' claim is not statistically supported. Please report standard deviations or per-split results and, if possible, paired significance tests.
  2. [Sec. IV-A and Tables VI-IX] Hyperparameters λ_retain, λ_base, N_coarse, N_neg, and the T_img/8 pooling size in Eq. 3 are selected via ablations on THUMOS'14 using the same test benchmark on which final numbers are reported. This makes the reported gains optimistic. Please either use a validation split for hyperparameter selection, or report sensitivity across the 10 random splits and state which choices were fixed before evaluation.
  3. [Sec. III-F, Eq. 5] The projection layer φ_proj is the only learned component of the coarse-to-fine classifier, and it is trained with C_novel set equal to C_base because only base annotations exist. At inference it must transfer to unseen novel categories. The paper provides no direct evidence for this transfer; Table IV only compares variants of the coarse/fine pipeline within the same category split, so it cannot detect transfer failure. Moreover, Eq. 5 writes S_np = F_np · F_coarse^T without showing φ_proj, so it is unclear whether F_np is the raw image feature or the projected feature. Please clarify the exact computation and add an experiment that measures base-to-novel transfer of φ_proj, e.g., train on a subset of base categories and evaluate on held-out base categories, or compare with and without the projection on novel categories.
  4. [Sec. III-D, Algorithm 1] The ground-truth generation pseudo-code is ambiguous. In the inner loop, for every temporal position i it appends one entry per annotation (either the ground-truth interval or (-1,-1)) to P_gt, so P_gt is a set of T_fpn × N_a entries, yet line 15 indexes P_gt_i as if it were a single interval associated with position i. This makes the action-presence training target undefined. Please correct the algorithm or replace it with a clear textual definition, e.g., assign the interval with maximum tIoU for overlapping cases.
  5. [Sec. IV-C] The ZS-TAL protocol is underspecified. The text says MGCA-Net 'removes the conventional classifier and action presence predictor and only uses the coarse-to-fine classifier,' but it does not describe how proposals are filtered or how thresholds are set in this mode. Since Table II is a central result, this inference protocol must be defined precisely for reproducibility.
minor comments (5)
  1. [Sec. III-F, Eq. 4-5] The dimensional relationship is unclear: F_np is defined as a mean of image features F_img, while F_coarse is selected from text features F_text. If D_img and D_text differ, Eq. 5 is not well-formed without the projection layer. Please state the dimensions explicitly and show where φ_proj is applied.
  2. [Sec. IV-F] The Limitations section acknowledges that unannotated action instances are treated as negatives during localizer training. This is a genuine limitation and should be discussed in relation to novel-category recall; please clarify whether the baseline methods share this issue and how it affects the comparison.
  3. [Sec. III-F, Eq. 3] The definition of coarse categories via 'top-N_coarse values' is imprecise when scores are tied or when fewer than N_coarse categories pass a meaningful threshold. Please specify tie handling and the minimum score threshold, if any.
  4. [Sec. IV-A] The paper reports averaging over 10 random splits but does not give the random seeds or the exact split generation procedure. This information is needed for reproducibility, especially since category splits drive the OV/ZS evaluation.
  5. [Fig. 3] The baseline description says the method with the conventional classifier, action presence predictor, and coarse-to-fine classifier removed 'uses a single template for action categories and predicts action categories based on the zero-shot capability of VLMs.' It is unclear whether proposals come from the localizer alone and how they are thresholded; please clarify or add a caption-level description.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: MGCA-Net's claims are evaluated against external benchmarks and its learned projection is trained on base categories while tested on unseen novel categories.

full rationale

The paper contains no derivation chain that reduces to its own inputs. The central OV-TAL/ZS-TAL claims are measured on external benchmarks (THUMOS'14 and ActivityNet-1.3) against published methods, using frozen CLIP features and the ActionFormer localizer as external components. The only learned component of the novel-category path, the projection layer φ_proj, is trained with C_novel set equal to C_base (Sec. III-F: 'we directly set the content of C_novel to be identical to C_base') and evaluated on disjoint novel categories at inference; this is a genuine base-to-novel generalization test, not a fitted prediction of the reported mAP novel. The coarse categories are generated in a training-free manner from frozen VLM similarities plus MIL, and the fine categories are assigned by matching proposal features to those coarse categories. No equation is shown to be identical to its training target by construction. The self-citations in the paper (e.g., refs. [1], [6], [7]) appear only in related-work enumerations of TAL methods and are not load-bearing for the architecture or the performance claims. The noted ambiguity in Eq. 5, where φ_proj is described in the text but not written in the similarity formula, is a notation omission rather than a circular step. Overall, the evaluation is self-contained against external benchmarks and the central contribution is not circular.

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

The central claim rests on empirical generalization assumptions: zero-shot VLM alignment, base-trained localizer coverage, and transfer of a base-only trained projection layer to novel categories. The reported performance also depends on several thresholds selected by ablations on the evaluation benchmarks.

free parameters (5)
  • λ_retain = 0.5
    Threshold on action presence score; ablation on THUMOS'14 selects 0.5 (Table VI).
  • λ_base = 0.5
    Threshold for base-category confidence; tuned in Table VII.
  • N_coarse = 2
    Number of coarse categories selected per video; ablation Table VIII shows optimum 2.
  • N_neg = 3
    Number of negative text categories in contrastive loss; ablation Table IX.
  • Top-image fraction T_img/8 = T_img/8
    MIL selects top T_img/8 image-frame similarities per category in Eq. 3; hand-chosen constant.
assumptions (4)
  • domain assumption CLIP image-text similarity transfers from images to video frames for action categories.
    The coarse-category and proposal-matching steps rely entirely on frozen CLIP features and MIL pooling (Sec. III-F).
  • domain assumption The ActionFormer localizer trained on base annotations produces category-agnostic proposals that cover novel actions.
    The localizer is trained only on base action instances; the paper's Limitations section acknowledges that unannotated instances treated as negatives can bias it.
  • domain assumption The projection layer φ_proj trained with C_novel=C_base transfers to unseen novel categories.
    Sec. III-F Loss Function trains the projection only against base categories because no novel labels exist at training time.
  • domain assumption Training annotations are complete, so all actions not labeled in the training set can be treated as background.
    The localizer and presence head use labels as ground truth; the paper's Limitations section notes this assumption is violated in practice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MGCA-Net: Multi-Grained Category-Aware Network for Open-Vocabulary Temporal Action Localization." pith.science (2026). https://pith.science/paper/Q36APAI3

@misc{pith2026251113039,
  author       = {Pith},
  title        = {Pith review of: MGCA-Net: Multi-Grained Category-Aware Network for Open-Vocabulary Temporal Action Localization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q36APAI3}},
  note         = {Machine review of arXiv:2511.13039}
}
read the original abstract

Open-Vocabulary Temporal Action Localization (OV-TAL) aims to recognize and localize instances of any desired action categories in videos without explicitly curating training data for all categories. Existing methods mostly recognize action categories at a single granularity, which degrades the recognition accuracy of both base and novel action categories. To address these issues, we propose a Multi-Grained Category-Aware Network (MGCA-Net) comprising a localizer, an action presence predictor, a conventional classifier, and a coarse-to-fine classifier. Specifically, the localizer localizes category-agnostic action proposals. For these action proposals, the action presence predictor estimates the probability that they belong to an action instance. At the same time, the conventional classifier predicts the probability of each action proposal over base action categories at the snippet granularity. Novel action categories are recognized by the coarse-to-fine classifier, which first identifies action presence at the video granularity. Finally, it assigns each action proposal to one category from the coarse categories at the proposal granularity. Through coarse-to-fine category awareness for novel actions and the conventional classifier's awareness of base actions, multi-grained category awareness is achieved, effectively enhancing localization performance. Comprehensive evaluations on the THUMOS'14 and ActivityNet-1.3 benchmarks demonstrate that our method achieves state-of-the-art performance. Furthermore, our MGCA-Net achieves state-of-the-art results under the Zero-Shot Temporal Action Localization setting.

Figures

Figures reproduced from arXiv: 2511.13039 by the authors.

Figure 1
Figure 1. The structural comparison of (a) the localization [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed MGCA-Net. MGCA-Net employs frozen video, image, and text encoders to extract video, [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visualization of the localization results produced by the baseline and our MGCA-Net on THUMOS’14, where the [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 5 linked inside Pith

  1. [1]

    Pcpcad: proposal complementary action detector,

    Z. Fang, S. Zhu, J. Yu, and Q. Tian, “Pcpcad: proposal complementary action detector,” in2019 IEEE International Conference on Multimedia and Expo (ICME). IEEE, 2019, pp. 424–429

    2019

  2. [2]

    Capsule boundary network with 3d convolutional dynamic routing for temporal action detection,

    Y . Chen, B. Guo, Y . Shen, W. Wang, W. Lu, and X. Suo, “Capsule boundary network with 3d convolutional dynamic routing for temporal action detection,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 32, no. 5, pp. 2962–2975, 2021

    2021

  3. [3]

    Actionformer: Localizing moments of actions with transformers,

    C.-L. Zhang, J. Wu, and Y . Li, “Actionformer: Localizing moments of actions with transformers,” inEuropean Conference on Computer Vision. Springer, 2022, pp. 492–510

    2022

  4. [4]

    Tridet: Temporal action detection with relative boundary modeling,

    D. Shi, Y . Zhong, Q. Cao, L. Ma, J. Li, and D. Tao, “Tridet: Temporal action detection with relative boundary modeling,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 18 857–18 866

    2023

  5. [5]

    Co- occurrence matters: Learning action relation for temporal action localiza- tion,

    C. Cao, Y . Wang, Y . Zhang, Y . Lu, X. Zhang, and Y . Zhang, “Co- occurrence matters: Learning action relation for temporal action localiza- tion,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 5, pp. 3327–3339, 2023

    2023

  6. [6]

    Lpr: learning point-level temporal action localization through re-training,

    Z. Fang, J. Fan, and J. Yu, “Lpr: learning point-level temporal action localization through re-training,”Multimedia Systems, vol. 29, no. 5, pp. 2545–2562, 2023

    2023

  7. [7]

    Boundary discretization and reliable classification network for temporal action detection,

    Z. Fang, J. Yu, and R. Hong, “Boundary discretization and reliable classification network for temporal action detection,”IEEE Transactions on Multimedia, 2025

    2025

  8. [8]

    Throughout procedural transformer for online action detection and anticipation,

    H. Yuan, Y . Chen, Z. Ji, Z. Zheng, Y . Gu, and J. Zhou, “Throughout procedural transformer for online action detection and anticipation,” IEEE Transactions on Circuits and Systems for Video Technology, 2025

    2025

Show all 49 references
  1. [9]

    Bfstal: Bidirectional feature splitting with cross-layer fusion for temporal action localization,

    J. Xu, Y . Zhang, W. Zhou, and H. Liu, “Bfstal: Bidirectional feature splitting with cross-layer fusion for temporal action localization,”IEEE Transactions on Circuits and Systems for Video Technology, 2025

    2025

  2. [10]

    Constructing semantical structure by segmentation integrated video embedding for temporal action detection,

    Z. Zhao, S. Liu, C. Zhao, and X. Zhao, “Constructing semantical structure by segmentation integrated video embedding for temporal action detection,”IEEE Transactions on Circuits and Systems for Video Technology, 2025

    2025

  3. [11]

    Opental: Towards open set temporal action localization,

    W. Bao, Q. Yu, and Y . Kong, “Opental: Towards open set temporal action localization,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 2979–2989

    2022

  4. [12]

    Ow-tal: learning unknown human activities for open-world temporal action localization,

    Y . Zhang, X.-Y . Zhang, and H. Shi, “Ow-tal: learning unknown human activities for open-world temporal action localization,”Pattern Recog- nition, vol. 133, p. 109027, 2023

    2023

  5. [13]

    Prompting visual- language models for efficient video understanding,

    C. Ju, T. Han, K. Zheng, Y . Zhang, and W. Xie, “Prompting visual- language models for efficient video understanding,” inEuropean con- ference on computer vision. Springer, 2022, pp. 105–124

    2022

  6. [14]

    Detal: Open- vocabulary temporal action localization with decoupled networks,

    Z. Li, Y . Zhong, R. Song, T. Li, L. Ma, and W. Zhang, “Detal: Open- vocabulary temporal action localization with decoupled networks,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 46, no. 12, pp. 7728–7741, 2024

    2024

  7. [15]

    Zeetad: Adapting pretrained vision-language model for zero-shot end-to-end temporal action detection,

    T. Phan, K. V o, D. Le, G. Doretto, D. Adjeroh, and N. Le, “Zeetad: Adapting pretrained vision-language model for zero-shot end-to-end temporal action detection,” inProceedings of the IEEE/CVF winter conference on applications of computer vision, 2024, pp. 7046–7055

    2024

  8. [16]

    Exploring scala- bility of self-training for open-vocabulary temporal action localization,

    J. Hyun, S. H. Han, H. Kang, J.-Y . Lee, and S. J. Kim, “Exploring scala- bility of self-training for open-vocabulary temporal action localization,” in2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). IEEE, 2025, pp. 9406–9415

    2025

  9. [17]

    Zero-shot temporal action detection via vision-language prompting,

    S. Nag, X. Zhu, Y .-Z. Song, and T. Xiang, “Zero-shot temporal action detection via vision-language prompting,” inEuropean conference on computer vision. Springer, 2022, pp. 681–697

    2022

  10. [18]

    Unloc: A unified framework for video localization tasks,

    S. Yan, X. Xiong, A. Nagrani, A. Arnab, Z. Wang, W. Ge, D. Ross, and C. Schmid, “Unloc: A unified framework for video localization tasks,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 13 623–13 633

    2023

  11. [19]

    Zero-shot temporal action detection by learning multimodal prompts and text-enhanced actionness,

    A. Raza, B. Yang, and Y . Zou, “Zero-shot temporal action detection by learning multimodal prompts and text-enhanced actionness,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 11, pp. 11 000–11 012, 2024

    2024

  12. [20]

    Text-infused attention and foreground-aware modeling for zero-shot temporal action detection,

    Y . Lee, H.-J. Kim, and S.-W. Lee, “Text-infused attention and foreground-aware modeling for zero-shot temporal action detection,” Advances in Neural Information Processing Systems, vol. 37, pp. 9864– 9884, 2024

    2024

  13. [21]

    Open-vocabulary temporal action localization using multimodal guidance,

    A. Gupta, A. Arora, S. Narayan, S. Khan, F. S. Khan, and G. W. Taylor, “Open-vocabulary temporal action localization using multimodal guidance,”arXiv preprint arXiv:2406.15556, 2024

    2024 arXiv

  14. [22]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clarket al., “Learning transferable visual models from natural language supervision,” inInternational conference on machine learning. PmLR, 2021, pp. 8748–8763

    2021

  15. [23]

    Scaling up visual and vision-language representation learning with noisy text supervision,

    C. Jia, Y . Yang, Y . Xia, Y .-T. Chen, Z. Parekh, H. Pham, Q. Le, Y .-H. Sung, Z. Li, and T. Duerig, “Scaling up visual and vision-language representation learning with noisy text supervision,” inInternational conference on machine learning. PMLR, 2021, pp. 4904–4916

    2021

  16. [24]

    Vltint: Visual-linguistic transformer-in-transformer for coherent video para- graph captioning,

    K. Yamazaki, K. V o, Q. S. Truong, B. Raj, and N. Le, “Vltint: Visual-linguistic transformer-in-transformer for coherent video para- graph captioning,” inProceedings of the AAAI Conference on Artificial intelligence, vol. 37, no. 3, 2023, pp. 3081–3090

    2023

  17. [25]

    Decoupling zero-shot semantic segmentation,

    J. Ding, N. Xue, G.-S. Xia, and D. Dai, “Decoupling zero-shot semantic segmentation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 11 583–11 592

    2022

  18. [26]

    Aisformer: Amodal instance segmentation with transformer,

    M. Tran, K. V o, K. Yamazaki, A. Fernandes, M. Kidd, and N. Le, “Aisformer: Amodal instance segmentation with transformer,”arXiv preprint arXiv:2210.06323, 2022

    2022 arXiv

  19. [27]

    The power of scale for parameter-efficient prompt tuning,

    B. Lester, R. Al-Rfou, and N. Constant, “The power of scale for parameter-efficient prompt tuning,”arXiv preprint arXiv:2104.08691, 2021

    2021 arXiv

  20. [28]

    R-c3d: Region convolutional 3d network for temporal activity detection,

    H. Xu, A. Das, and K. Saenko, “R-c3d: Region convolutional 3d network for temporal activity detection,” inProceedings of the IEEE international conference on computer vision, 2017, pp. 5783–5792

    2017

  21. [29]

    G-tad: Sub- graph localization for temporal action detection,

    M. Xu, C. Zhao, D. S. Rojas, A. Thabet, and B. Ghanem, “G-tad: Sub- graph localization for temporal action detection,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 10 156–10 165

    2020

  22. [30]

    Tallformer: Temporal action localization with a long-memory transformer,

    F. Cheng and G. Bertasius, “Tallformer: Temporal action localization with a long-memory transformer,” inEuropean Conference on Computer Vision. Springer, 2022, pp. 503–521

    2022

  23. [31]

    End- to-end temporal action detection with transformer,

    X. Liu, Q. Wang, Y . Hu, X. Tang, S. Zhang, S. Bai, and X. Bai, “End- to-end temporal action detection with transformer,”IEEE Transactions on Image Processing, vol. 31, pp. 5427–5441, 2022

    2022

  24. [32]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,”Advances in neural information processing systems, vol. 30, 2017

    2017

  25. [33]

    End-to-end object detection with transformers,

    N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, and S. Zagoruyko, “End-to-end object detection with transformers,” in European conference on computer vision. Springer, 2020, pp. 213– 229

    2020

  26. [34]

    Weakly supervised action localization by sparse temporal pooling network,

    P. Nguyen, T. Liu, G. Prasad, and B. Han, “Weakly supervised action localization by sparse temporal pooling network,” inProceedings of the JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 12 IEEE conference on computer vision and pattern recognition, 2018, pp. 6752–6761

    2021

  27. [35]

    Pivotal: Prior-driven supervision for weakly-supervised temporal action localization,

    M. N. Rizve, G. Mittal, Y . Yu, M. Hall, S. Sajeev, M. Shah, and M. Chen, “Pivotal: Prior-driven supervision for weakly-supervised temporal action localization,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 22 992–23 002

    2023

  28. [36]

    Improving weakly supervised temporal action localization by bridging train-test gap in pseudo labels,

    J. Zhou, L. Huang, L. Wang, S. Liu, and H. Li, “Improving weakly supervised temporal action localization by bridging train-test gap in pseudo labels,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 23 003–23 012

    2023

  29. [37]

    Learning action completeness from points for weakly-supervised temporal action localization,

    P. Lee and H. Byun, “Learning action completeness from points for weakly-supervised temporal action localization,” inProceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 13 648–13 657

    2021

  30. [38]

    Hr-pro: Point-supervised temporal action localization via hierarchical reliability propagation,

    H. Zhang, X. Wang, X. Xu, Z. Qing, C. Gao, and N. Sang, “Hr-pro: Point-supervised temporal action localization via hierarchical reliability propagation,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 7, 2024, pp. 7115–7123

    2024

  31. [39]

    Stepwise multi-grained boundary detector for point-supervised tempo- ral action localization,

    M. Liu, L. Wang, S. Zhou, K. Xia, Q. Wu, Q. Zhang, and G. Hua, “Stepwise multi-grained boundary detector for point-supervised tempo- ral action localization,” inEuropean Conference on Computer Vision. Springer, 2024, pp. 333–349

    2024

  32. [40]

    Distance-iou loss: Faster and better learning for bounding box regression,

    Z. Zheng, P. Wang, W. Liu, J. Li, R. Ye, and D. Ren, “Distance-iou loss: Faster and better learning for bounding box regression,” inProceedings of the AAAI conference on artificial intelligence, vol. 34, no. 07, 2020, pp. 12 993–13 000

    2020

  33. [41]

    Focal loss for dense object detection,

    T.-Y . Lin, P. Goyal, R. Girshick, K. He, and P. Doll ´ar, “Focal loss for dense object detection,” inProceedings of the IEEE international conference on computer vision, 2017, pp. 2980–2988

    2017

  34. [42]

    Open-vocabulary object detection via vision and language knowledge distillation,

    X. Gu, T.-Y . Lin, W. Kuo, and Y . Cui, “Open-vocabulary object detection via vision and language knowledge distillation,” inInternational Conference on Learning Representations, 2022. [Online]. Available: https://openreview.net/forum?id=lL3lnMbR4WU

    2022

  35. [43]

    Ov-dquo: Open-vocabulary detr with denoising text query training and open-world unknown objects supervision,

    J. Wang, B. Chen, B. Kang, Y . Li, W. Xian, Y . Chen, and Y . Xu, “Ov-dquo: Open-vocabulary detr with denoising text query training and open-world unknown objects supervision,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 7, 2025, pp. 7762– 7770

    2025

  36. [44]

    The thumos challenge on action recognition for videos “in the wild

    H. Idrees, A. R. Zamir, Y .-G. Jiang, A. Gorban, I. Laptev, R. Sukthankar, and M. Shah, “The thumos challenge on action recognition for videos “in the wild”,”Computer Vision and Image Understanding, vol. 155, pp. 1–23, 2017

    2017

  37. [45]

    Activitynet: A large-scale video benchmark for human activity under- standing,

    F. Caba Heilbron, V . Escorcia, B. Ghanem, and J. Carlos Niebles, “Activitynet: A large-scale video benchmark for human activity under- standing,” inProceedings of the ieee conference on computer vision and pattern recognition, 2015, pp. 961–970

    2015

  38. [46]

    Quo vadis, action recognition? a new model and the kinetics dataset,

    J. Carreira and A. Zisserman, “Quo vadis, action recognition? a new model and the kinetics dataset,” inproceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 6299–6308

    2017

  39. [47]

    Tsp: Temporally-sensitive pretraining of video encoders for localization tasks,

    H. Alwassel, S. Giancola, and B. Ghanem, “Tsp: Temporally-sensitive pretraining of video encoders for localization tasks,” inProceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 3173–3183

    2021

  40. [48]

    The kinetics human action video dataset,

    W. Kay, J. Carreira, K. Simonyan, B. Zhang, C. Hillier, S. Vijaya- narasimhan, F. Viola, T. Green, T. Back, P. Natsevet al., “The kinetics human action video dataset,”arXiv preprint arXiv:1705.06950, 2017

    2017 arXiv

  41. [49]

    Sgdr: Stochastic gradient descent with warm restarts,

    I. Loshchilov and F. Hutter, “Sgdr: Stochastic gradient descent with warm restarts,”arXiv preprint arXiv:1608.03983, 2016

    2016 arXiv

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.