Pith. sign in

REVIEW 3 major objections 4 minor 64 references

TAMT: Temporal-Aware Model Tuning for Cross-Domain Few-Shot Action Recognition

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper claims that cross-domain few-shot action recognition can be solved by pre-training once on the source and fine-tuning only a handful of temporal-aware parameters on the target, beating joint-training approaches by 13–31% at a…

desk verdict TAMT is a sensible parameter-efficient tuning recipe for CDFSAR, but the headline 13–31% margin is mostly a pre-training and resolution effect; the method's own matched-setting gains are a modest 2–3 points. read the letter →

arxiv 2411.19041 v2 pith:S2RL7W2E submitted 2024-11-28 cs.CV

Yilong Wang , Zilin Gao , Qilong Wang , Zhaofeng Chen , Peihua Li , Qinghua Hu This is my paper · ORCID
classification cs.CV
keywords cross-domainfew-shotactionrecognitiontemporaladaptermomentpoolingparameter-efficientfine-tuningVideoMAEself-supervisedpre-trainingtransferlearningprototypicalnetwork
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

Cross-domain few-shot action recognition normally trains source and target data together, so every new target means retraining the whole model. This paper tries to establish that a decoupled pipeline works better: pre-train the video model once on the source data, freeze it, and adapt to the target with a small set of temporal-aware modules. The proposed TAMT baseline reports accuracy gains of 13–31% over joint-training state-of-the-art methods across five source and five target benchmarks, while using far less training compute. The central claim is that a frozen spatio-temporal backbone plus lightweight local adapters and moment-based global pooling is enough to close the cross-domain gap.

What carries the argument

The load-bearing object is HTTN, composed of Temporal-Aware Adapters (TAA) and Global Temporal-aware Moment Tuning (GTMT). TAA computes a per-frame temporal scale $\gamma$ and bias $\beta$ from the pooled features of a frozen transformer block, using a shared two-layer temporal convolution with a bottleneck ratio of 4 and kernel size 3; these scale and shift parameters recalibrate intermediate features with only 2.8 million trainable parameters. GTMT produces the video representation from two moments: the global-average-pooled first moment and a second moment computed by the Efficient Long-Short Temporal Covariance layer (ELSTC), which splits the frame sequence into groups, computes long-short temporal covariance matrices inside each group, and aggregates them with convolutions. The two moments are combined and matched by Euclidean distance in a prototypical network. This object carries the argument because it is the only component tuned on the target data; the backbone stays frozen.

What would settle it

Train CDFSL-V, SEEN, or DMSD under exactly the same protocol as TAMT—same VideoMAE SSL+SL pre-trained checkpoint at 112×112 resolution, same number of fine-tuning epochs, same prototypical evaluation—and compare. If the margin collapses, the claimed advantage is not caused by TAA and GTMT but by the pre-training recipe or resolution.

Watch

Extended reading notes

Core claim

The central discovery claimed is that a VideoMAE backbone pre-trained on source data with self-supervised reconstruction followed by supervised classification can be frozen and transferred to an unseen target domain using only about 2.8 million trainable parameters. The transfer is done by the Hierarchical Temporal Tuning Network (HTTN): local Temporal-Aware Adapters rescale and shift frame-level features inside the last transformer blocks, while Global Temporal-aware Moment Tuning pools the resulting features into a representation built from first-order and second-order temporal covariance statistics. With a prototypical nearest-neighbour classifier, this beats full fine-tuning and outperforms prior joint-training CDFSAR methods by 13–31% on the reported 5-way 5-shot benchmarks, for example 68.37% average accuracy with K-400 as the source at 112×112 resolution.

Load-bearing premise

The 13–31% advantage over earlier methods is caused by the temporal-aware tuning modules rather than by the stronger two-stage pre-training recipe (self-supervised then supervised) and different input resolution that those earlier methods did not receive.

Editorial extensions

If this is right

  • One source pre-training serves many target datasets: with K-400 as source, TAMT is reported to use 19 GPU days across five targets versus CDFSL-V's 88.
  • Under the same SSL+SL pre-training, TAMT beats full fine-tuning by about 1–3% on average while using roughly one tenth of the GPU memory and one tenth of the trainable parameters.
  • The gains extend to other shot settings: with K-100 as source, TAMT reports average improvements over CDFSL-V of 24.08% at 1-shot and 34.13% at 20-shot.
  • After tuning on UCF or HMDB, the model transfers to four unseen target datasets without further tuning, beating CDFSL-V by 13.83% and 15.26% on average.
  • The same HTTN tuning also works for standard few-shot action recognition with a ViT-B backbone, matching or exceeding full fine-tuning of CLIP-based methods.

Reading between the lines

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

  • Because the paper's own ablation shows SSL+SL pre-training lifts full fine-tuning from 54.67 to 64.30 average, a matched-recipe comparison with earlier baselines would isolate how much of the headline margin comes from the HTTN modules rather than from the pre-training recipe.
  • Because TAA and GTMT are plug-in modules on a frozen encoder, they could be reused for other video transfer tasks such as temporal action localization or video retrieval with little modification.
  • The ELSTC grouping trick suggests that the temporal-covariance representation could be sharpened further with learnable temporal grouping, a direction the paper does not explore.
  • The decoupled paradigm turns cross-domain few-shot action recognition into a parameter-efficient transfer problem, so the gains should compound when the frozen source model is a larger video foundation model.
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

3 major / 4 minor

Summary. The paper proposes TAMT, a decoupled pre-training/fine-tuning approach for cross-domain few-shot action recognition (CDFSAR). It freezes a VideoMAE backbone and introduces Temporal-Aware Adapters (TAA) into the last transformer blocks plus a Global Temporal-aware Moment Tuning (GTMT) module with Efficient Long-Short Temporal Covariance (ELSTC) for video representation. Experiments across five source and five target datasets report 5-way 5-shot accuracy gains of 13% to 31% over prior CDFSAR methods, along with reduced training memory and a claim of 19 GPU days versus 88 GPU days for CDFSL-V when training on K-400 for five targets.

Significance. If the reported margins hold under controlled comparison, TAMT would offer a valuable parameter-efficient baseline for CDFSAR, and the decoupled paradigm would reduce re-training cost when a single source is paired with multiple targets. The paper also provides ablations showing that TAA and GTMT each contribute under a fixed SSL+SL recipe. However, the headline state-of-the-art comparison does not isolate the proposed modules from the stronger SSL+SL pre-training recipe, nor does it match prior methods in backbone and resolution; the central claim therefore currently rests on an attribution that is not fully supported by the experiments as presented.

major comments (3)
  1. [§4.2, Table 1 and Table 2] The headline 13%–31% comparison is not controlled for the pre-training recipe. Table 2 shows that for the K-400 source with ViT-S, moving from SSL-only to SSL+SL pre-training raises the FFT baseline average on SSV2/Diving/UCF from 55.01 to 64.30, and TAMT from 52.80 to 66.76; TAMT is only +2.46 over FFT under the same SSL+SL recipe. Since prior methods in Table 1 (e.g., CDFSL-V average 47.24 on K-400 across five targets) are not trained with the same SSL+SL VideoMAE recipe, the pre-training recipe alone can account for a substantial portion of the reported gains. The authors should retrain at least the main baselines under the same recipe, backbone, and resolution, or explicitly restrict the claim to beating the published numbers under the original protocols.
  2. [§4.2, Table 1] The comparison with SEEN and DMSD is confounded by resolution and backbone: SEEN and DMSD are marked as running at 224×224 resolution with ResNet-18, while the unstarred TAMT runs at 112×112 with ViT-S. Because input resolution and backbone have a large effect on few-shot video accuracy, the reported outperformance over SEEN and DMSD cannot be attributed to the proposed modules from this table alone. Please provide matched-protocol comparisons (same backbone and resolution for all methods, or both configurations for TAMT) or clearly state that the comparison is cross-protocol.
  3. [§4.3, Table 4(a) and Table S5] The ablation margins used to justify specific design choices are small: ELSTC over TCov gives +0.75 average accuracy, GTMT over ELSTC gives +0.82, and L=2 versus L=3 differs by −0.01, yet no confidence intervals or multi-seed statistics are reported despite averaging 10,000 episodes. These differences are load-bearing for the claims that ELSTC is preferable to TCov and that adding the first-order moment is beneficial; without variance estimates, the reader cannot assess whether these differences are meaningful. Please add confidence intervals and/or results over multiple seeds.
minor comments (4)
  1. [§3.3, Eq. (4) and Eq. (10)] The characteristic function expansion motivates the use of moments, but the final representation in Eq. (10) is a linear combination H(M2) ⊕ M1 that does not use the coefficients α_p from Eq. (4). Consider presenting Eq. (4) as a heuristic motivation and explicitly stating that the aggregation is a learned linear combination rather than an evaluation of the characteristic function.
  2. [References, [33]] Reference [33] lists the venue as 'Advances in Neural Information Processing Systems (ICCV)'; the correct venue for ST-Adapter is NeurIPS.
  3. [§3.3, ELSTC description] The phrase 'the timestamps t and t′ always have a temporal gap Δ, ranging from 0 to (T−G) with an interval of G' is confusing, because grouping along the temporal dimension changes the indexing of frames within each group. Please define t and t′ clearly with respect to the grouped feature X_ge.
  4. [Table 1 caption and Fig. 1b] The asterisk on TAMT* in some blocks is used for the 224×224 version, but in the K-100 block the same marker appears on SEEN*† and DMSD*† as well; the caption should state explicitly which entries are at 224×224 versus 112×112 and which use ResNet-18.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: TAMT's gains are external benchmark comparisons; the self-cited moment-probing paper is motivational only, and the SSL+SL attribution caveat is a fairness issue, not an equation-to-input reduction.

full rationale

The paper's derivation chain is empirical: the method pre-trains a VideoMAE encoder on source data, then tunes TAA/GTMT on target support sets, and reports external benchmark accuracies in Table 1. Equations (1)-(10) define the modules; no reported accuracy is the value of a fitted parameter, and no prediction reduces by construction to an input. Hyperparameters (L, G, rho, tau, CM) are set by ablations in the paper/supplement, which is selection risk, not circularity. The only overlapping-author citation is [11] (moment probing), used to motivate Eq. (4); it is not load-bearing because the module is explicitly defined and Table 4 compares it against GAP, TCov, Adapter, SSF, TAM, and ST-Adapter. The main validity caveat is fairness: Table 2 shows SSL+SL pre-training alone lifts the FFT baseline on SSV2/Diving/UCF from 54.67 to 64.30 (TAMT: 66.76), and Table 1 comparators include methods without that recipe and at 224x224/ResNet-18 (SEEN*, DMSD*); footnote 2 even concedes FFT beats TAMT under SSL-only. Those are attribution issues, not circularity, so the score stays 0.

Assumptions & free parameters 7 free parameters · 3 assumptions · 0 invented entities

The central claim rests on benchmark comparisons rather than on fitted constants that encode the result. The hyperparameters L, G, tau, C_M, rho, k_t, and k_c are chosen by hand or ablation, so they carry selection risk but do not make the result circular. No new particles, forces, or physical entities are introduced.

free parameters (7)
  • L (number of TAA blocks) = 2
    Selected by ablation over L=0..4 in Tab. S5; affects accuracy and memory.
  • G (number of ELSTC temporal groups) = 4
    Selected by ablation over G=1..8 in Tab. S7; balances efficiency and accuracy.
  • tau (ELSTC channel reduction factor) = 6
    Fixed in Sec. 3.3; controls covariance dimension and computation.
  • C_M (output covariance dimension) = 64
    Fixed in Sec. 3.3; sets final second-order representation size.
  • rho (TAA bottleneck reduction factor) = 4
    Fixed in Sec. 3.3; controls TAA parameter count.
  • k_t (temporal convolution kernel in TAA) = 3
    Fixed in Sec. 3.3; captures local temporal context.
  • k_c (aggregation convolution kernel in GTMT) = 3
    Fixed in Sec. 3.3; used in the two-layer aggregation A.
assumptions (3)
  • domain assumption Frozen VideoMAE features plus a small number of tunable adapters are sufficient to adapt a source pre-trained model to a target video domain.
    This is the central premise of the parameter-efficient tuning design; it is supported by ablations but not derived.
  • domain assumption Source and target classes are disjoint and the benchmark splits follow the CDFSL-V protocol.
    Stated in Sec. 3.1 and Sec. 4.1; required for the cross-domain few-shot evaluation to be meaningful.
  • ad hoc to paper First- and second-order moments of spatio-temporal features approximate the feature distribution used for matching, as written in Eq. (4).
    The characteristic-function expansion is informal: it omits u-dependent coefficients and imaginary units. The paper uses this as motivation for GTMT rather than as a proved identity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TAMT: Temporal-Aware Model Tuning for Cross-Domain Few-Shot Action Recognition." pith.science (2026). https://pith.science/paper/S2RL7W2E

@misc{pith2026241119041,
  author       = {Pith},
  title        = {Pith review of: TAMT: Temporal-Aware Model Tuning for Cross-Domain Few-Shot Action Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S2RL7W2E}},
  note         = {Machine review of arXiv:2411.19041}
}
abstract

Going beyond few-shot action recognition (FSAR), cross-domain FSAR (CDFSAR) has attracted recent research interests by solving the domain gap lying in source-to-target transfer learning. Existing CDFSAR methods mainly focus on joint training of source and target data to mitigate the side effect of domain gap. However, such kind of methods suffer from two limitations: First, pair-wise joint training requires retraining deep models in case of one source data and multiple target ones, which incurs heavy computation cost, especially for large source and small target data. Second, pre-trained models after joint training are adopted to target domain in a straightforward manner, hardly taking full potential of pre-trained models and then limiting recognition performance. To overcome above limitations, this paper proposes a simple yet effective baseline, namely Temporal-Aware Model Tuning (TAMT) for CDFSAR. Specifically, our TAMT involves a decoupled paradigm by performing pre-training on source data and fine-tuning target data, which avoids retraining for multiple target data with single source. To effectively and efficiently explore the potential of pre-trained models in transferring to target domain, our TAMT proposes a Hierarchical Temporal Tuning Network (HTTN), whose core involves local temporal-aware adapters (TAA) and a global temporal-aware moment tuning (GTMT). Particularly, TAA learns few parameters to recalibrate the intermediate features of frozen pre-trained models, enabling efficient adaptation to target domains. Furthermore, GTMT helps to generate powerful video representations, improving match performance on the target domain. Experiments on several widely used video benchmarks show our TAMT outperforms the recently proposed counterparts by 13%$\sim$31%, achieving new state-of-the-art CDFSAR results.

Figures

Figures reproduced from arXiv: 2411.19041 by the authors.

Figure 1
Figure 1. (a) Comparison of existing CDFSAR methods in terms of training paradigm under the case of a single source data [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. (a) Overview of our TAMT paradigm, which pre-trains the models on [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Overview of our proposed Hierarchical Temporal Tuning Network (HTTN), where (a) local temporal-aware adapters (TAA) are [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Convergence curves of FFT and TAMT on SSV2 dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 59 canonical work pages

  1. [1]

    StyleDomain: Efficient and lightweight parameteriza- tions of StyleGAN for one-shot and few-shot domain adap- tation

    Aibek Alanov, Vadim Titov, Maksim Nakhodnov, and Dmitry Vetrov. StyleDomain: Efficient and lightweight parameteriza- tions of StyleGAN for one-shot and few-shot domain adap- tation. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision (ICCV), pages 2184–2194, 2023. 1

    work page 2023

  2. [2]

    ViViT: A video vision transformer

    Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lucic, and Cordelia Schmid. ViViT: A video vision transformer. In IEEE/CVF International Conference on Computer Vision (ICCV), pages 6816–6826, 2021. 2

    work page 2021

  3. [3]

    TARN: Temporal attentive relation network for few-shot and zero-shot action recognition

    Mina Bishay, Georgios Zoumpourlis, and Ioannis Patras. TARN: Temporal attentive relation network for few-shot and zero-shot action recognition. British Machine Vision Confer- ence (BMVC), pages 130.1–130.14, 2019. 2

    work page 2019

  4. [4]

    Pattern recognition and machine learning

    Christopher M Bishop and Nasser M Nasrabadi. Pattern recognition and machine learning. Springer, 2006. 5

    work page 2006

  5. [5]

    Few-shot action recognition with implicit temporal alignment and pair similarity optimization

    Congqi Cao, Yajuan Li, Qinyi Lv, Peng Wang, and Yanning Zhang. Few-shot action recognition with implicit temporal alignment and pair similarity optimization. Computer Vision and Image Understanding (CVIU), 210:103250, 2021. 2

    work page 2021

  6. [6]

    Few-shot video classification via tempo- ral alignment

    Kaidi Cao, Jingwei Ji, Zhangjie Cao, Chien Yi Chang, and Juan Carlos Niebles. Few-shot video classification via tempo- ral alignment. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10615–10624, 2020. 1, 2, 4

    work page 2020

  7. [7]

    Quo vadis, Action Recognition? A new model and the kinetics dataset

    Joao Carreira and Andrew Zisserman. Quo vadis, Action Recognition? A new model and the kinetics dataset. In proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 6299–6308, 2017. 2, 6

    work page 2017

  8. [8]

    Tem- adapter: Adapting image-text pretraining for video question answer

    Guangyi Chen, Xiao Liu, Guangrun Wang, Kun Zhang, Philip HS Torr, Xiao-Ping Zhang, and Yansong Tang. Tem- adapter: Adapting image-text pretraining for video question answer. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision (ICCV) , pages 13945–13955,

    work page

Show all 64 references
  1. [9]

    Semantic segmentation on VSPW dataset through aggregation of trans- former models

    Zixuan Chen, Junhong Zou, and Xiaotao Wang. Semantic segmentation on VSPW dataset through aggregation of trans- former models. arXiv preprint arXiv:2109.01316, 2021. 2

    2021 arXiv

  2. [10]

    ImageNet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. ImageNet: A large-scale hierarchical image database. In IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR), pages 248–255, 2009. 2

    2009

  3. [11]

    Tuning pre-trained model via moment probing

    Mingze Gao, Qilong Wang, Zhenyi Lin, Pengfei Zhu, Qinghua Hu, and Jingbo Zhou. Tuning pre-trained model via moment probing. In Proceedings of the IEEE/CVF In- ternational Conference on Computer Vision (ICCV), pages 11803–11813, 2023. 5

    2023

  4. [12]

    Exploring the cross-domain action recognition prob- lem by deep feature learning and cross-domain learning.IEEE Access, 6:68989–69008, 2018

    Zan Gao, Tao-tao Han, Lei Zhu, Hua Zhang, and Yinglong Wang. Exploring the cross-domain action recognition prob- lem by deep feature learning and cross-domain learning.IEEE Access, 6:68989–69008, 2018. 3

    2018

  5. [13]

    A pairwise attentive adversarial spatiotemporal network for cross-domain few-shot action recognition-R2

    Zan Gao, Leming Guo, Weili Guan, Anan Liu, and Shengy- ong Chen. A pairwise attentive adversarial spatiotemporal network for cross-domain few-shot action recognition-R2. IEEE Transactions on Image Processing, 30:767–782, 2020. 3

    2020

  6. [14]

    A novel multiple-view adversarial learning network for unsupervised domain adaptation action recog- nition

    Zan Gao, Yibo Zhao, Hua Zhang, Da Chen, An-An Liu, and Shengyong Chen. A novel multiple-view adversarial learning network for unsupervised domain adaptation action recog- nition. IEEE Transactions on Cybernetics , 52(12):13197– 13211, 2022. 3

    2022

  7. [15]

    Towards better robustness against common corruptions for unsupervised domain adaptation

    Zhiqiang Gao, Kaizhu Huang, Rui Zhang, Dawei Liu, and Jieming Ma. Towards better robustness against common corruptions for unsupervised domain adaptation. In IEEE International Conference on Computer Vision (ICCV), pages 18882–18893, 2023. 1

    2023

  8. [16]

    Something Something

    Raghav Goyal, Samira Ebrahimi Kahou, Vincent Michal- ski, Joanna Materzynska, Susanne Westphal, Heuna Kim, Valentin Haenel, Ingo Fruend, Peter Yianilos, Moritz Mueller- Freitag, et al. The “Something Something” video database for learning and evaluating visual common sense. In...

    2017

  9. [17]

    Con- sistency prototype module and motion compensation for few- shot action recognition (CLIP-CPM 2C)

    Fei Guo, YiKang Wang, Han Qi, Li Zhu, and Jing Sun. Con- sistency prototype module and motion compensation for few- shot action recognition (CLIP-CPM 2C). Neurocomputing, 611:128649, 2025. 1, 4

    2025

  10. [18]

    DMSD-CDFSAR: Distillation from mixed-source domain for cross-domain few-shot action recognition

    Fei Guo, Yi Kang Wang, Han Qi, Li Zhu, and Jing Sun. DMSD-CDFSAR: Distillation from mixed-source domain for cross-domain few-shot action recognition. Expert Systems With Applications, 270, 2025. 3, 7

    2025

  11. [19]

    Codella, Leonid Karlinsky, James V

    Yunhui Guo, Noel C. Codella, Leonid Karlinsky, James V . Codella, and Rogerio Feris. A broader study of cross-domain few-shot learning. In European Conference on Computer Vision (ECCV), pages 124–141, 2020. 3

    2020

  12. [20]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll´ar, and Ross Girshick. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 16000–16009, 2022. 3

    2022

  13. [21]

    Parameter-efficient transfer learning for NLP

    Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin De Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. Parameter-efficient transfer learning for NLP. In International Conference on Machine Learning (ICML), pages 2790–2799, 2019. 4, 5, 8

    2019

  14. [22]

    Dy- namic distillation network for cross-domain few-shot recog- nition with unlabeled data

    Ashraful Islam, Chun-Fu Richard Chen, Rameswar Panda, Leonid Karlinsky, Rogerio Feris, and Richard J Radke. Dy- namic distillation network for cross-domain few-shot recog- nition with unlabeled data. Advances in Neural Information Processing Systems (NeurIPS), 34:3584–3595, 2021. 7, 3

    2021

  15. [23]

    Kuehne, H

    H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre. HMDB: A large video database for human motion recogni- tion. In International Conference on Computer Vision (ICCV), pages 2556–2563, 2011. 2, 6

    2011

  16. [24]

    UniFormerV2: Unlocking the 9 potential of image vits for video understanding

    Kunchang Li, Yali Wang, Yinan He, Yizhuo Li, Yi Wang, Limin Wang, and Yu Qiao. UniFormerV2: Unlocking the 9 potential of image vits for video understanding. In Proceed- ings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 1632–1643, 2023. 3

    2023

  17. [25]

    RESOUND: To- wards action recognition without representation bias

    Yingwei Li, Yi Li, and Nuno Vasconcelos. RESOUND: To- wards action recognition without representation bias. In Pro- ceedings of the European Conference on Computer Vision (ECCV), pages 513–528, 2018. 2, 6

    2018

  18. [26]

    Scaling & shifting your features: A new baseline for efficient model tuning

    Dongze Lian, Daquan Zhou, Jiashi Feng, and Xinchao Wang. Scaling & shifting your features: A new baseline for efficient model tuning. In Advances in Neural Information Processing Systems (NeurIPS), pages 109–123, 2022. 4, 5, 8

    2022

  19. [27]

    Deep quality assessment of com- pressed videos: A subjective and objective study.IEEE Trans- actions on Circuits and Systems for Video Technology, 33(6): 2616–2626, 2023

    Liqun Lin, Zheng Wang, Jiachen He, Weiling Chen, Yiwen Xu, and Tiesong Zhao. Deep quality assessment of com- pressed videos: A subjective and objective study.IEEE Trans- actions on Circuits and Systems for Video Technology, 33(6): 2616–2626, 2023. 2

    2023

  20. [28]

    MASTAF: A model-agnostic spatio-temporal attention fusion network for few-shot video classification

    Xin Liu, Huanle Zhang, Hamed Pirsiavash, and Xin Liu. MASTAF: A model-agnostic spatio-temporal attention fusion network for few-shot video classification. In Proceedings of the IEEE/CVF Winter Conference on Applications of Com- puter Vision (WACV), pages 2508–2517, 2023. 1, 3, 6, 4

    2023

  21. [29]

    TAM: Temporal adaptive module for video recognition

    Zhaoyang Liu, Limin Wang, Wayne Wu, Chen Qian, and Tong Lu. TAM: Temporal adaptive module for video recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 13708–13718, 2021. 8

    2021

  22. [30]

    Fine-grained unsupervised do- main adaptation for gait recognition

    Kang Ma, Ying Fu, Dezhi Zheng, Yunjie Peng, Chunshui Cao, and Yongzhen Huang. Fine-grained unsupervised do- main adaptation for gait recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 11313–11322, 2023. 1

    2023

  23. [31]

    RareAct: A video dataset of unusual interactions

    Antoine Miech, Jean Baptiste Alayrac, Ivan Laptev, Josef Sivic, and Andrew Zisserman. RareAct: A video dataset of unusual interactions. arXiv preprint arXiv:2008.01018, 2020. 2, 6

    2008 arXiv

  24. [32]

    DINOv2: Learning robust visual features without supervision

    Maxime Oquab, Timoth´ee Darcet, Th´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. DINOv2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193, 2023. 3

    2023 arXiv

  25. [33]

    ST-Adapter: Parameter-efficient image-to-video transfer learning

    Junting Pan, Ziyi Lin, Xiatian Zhu, Jing Shao, and Hongsheng Li. ST-Adapter: Parameter-efficient image-to-video transfer learning. Advances in Neural Information Processing Systems (ICCV), 35:26462–26477, 2022. 5, 8

    2022

  26. [34]

    Temporal-relational cross Transformers for few-shot action recognition

    Toby Perrett, Alessandro Masullo, Tilo Burghardt, Majid Mirmehdi, and Dima Damen. Temporal-relational cross Transformers for few-shot action recognition. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 475–484, 2021. 2, 4

    2021

  27. [35]

    Self-training for few-shot transfer across extreme task differences

    Cheng Perng Phoo and Bharath Hariharan. Self-training for few-shot transfer across extreme task differences. In Interna- tional Conference on Learning Representations (ICLR), 2021. 7, 3

    2021

  28. [36]

    A. J. Piergiovanni, Weicheng Kuo, and Anelia Angelova. Rethinking Video ViTs: Sparse video tubes for joint image and video learning. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 2214–2224,

  29. [37]

    Contrast with reconstruct: Contrastive 3D representation learning guided by generative pretraining

    Zekun Qi, Runpei Dong, Guofan Fan, Zheng Ge, Xiangyu Zhang, Kaisheng Ma, and Li Yi. Contrast with reconstruct: Contrastive 3D representation learning guided by generative pretraining. In International Conference on Machine Learn- ing (ICML), 2023. 3

    2023

  30. [38]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, and Jack Clark. Learning transferable visual models from natural language supervision. In International Conference on Machine Learning (ICML)...

    2021

  31. [39]

    CDFSL-V: Cross-domain few-shot learning for videos

    Sarinda Samarasinghe, Mamshad Nayeem Rizve, Navid Kar- dan, and Mubarak Shah. CDFSL-V: Cross-domain few-shot learning for videos. In Proceedings of the IEEE/CVF in- ternational conference on computer vision (ICCV) , pages 11643–11652, 2023. 1, 3, 5, 6, 7, 8, 2, 4

    2023

  32. [40]

    Com- monsense knowledge prompting for few-shot action recogni- tion in videos

    Yuheng Shi, Xinxiao Wu, Hanxi Lin, and Jiebo Luo. Com- monsense knowledge prompting for few-shot action recogni- tion in videos. IEEE Transactions on Multimedia, 26:8395– 8405, 2024. 6

    2024

  33. [41]

    Prototypical networks for few-shot learning

    Jake Snell, Kevin Swersky, and Richard Zemel. Prototypical networks for few-shot learning. In Proceedings of the 31st International Conference on Neural Information Processing Systems (NeurIPS), pages 4080–4090, 2017. 4

    2017

  34. [42]

    UCF101: A dataset of 101 human actions classes from videos in the wild

    Khurram Soomro, Amir Roshan Zamir, and Mubarak Shah. UCF101: A dataset of 101 human actions classes from videos in the wild. arXiv preprint arXiv:1212.0402, 2012. 2, 6

    2012 arXiv

  35. [43]

    Spatio-temporal relation modeling for few-shot action recognition

    Anirudh Thatipelli, Sanath Narayan, Salman Khan, Rao Muhammad Anwer, Fahad Shahbaz Khan, and Bernard Ghanem. Spatio-temporal relation modeling for few-shot action recognition. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 19926–19935,

  36. [44]

    Video- MAE: Masked autoencoders are data-efficient learners for self-supervised video pre-training

    Zhan Tong, Yibing Song, Jue Wang, and Limin Wang. Video- MAE: Masked autoencoders are data-efficient learners for self-supervised video pre-training. Advances in neural in- formation processing systems (NeurIPS), 35:10078–10093,

  37. [45]

    VideoMAE V2: Scaling video masked autoencoders with dual masking

    Limin Wang, Bingkun Huang, Zhiyu Zhao, Zhan Tong, Yi- nan He, Yi Wang, Yali Wang, and Yu Qiao. VideoMAE V2: Scaling video masked autoencoders with dual masking. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition (CVPR), pages 14549–14560,

  38. [46]

    Hybrid relation guided set matching for few-shot action recog- nition

    Xiang Wang, Shiwei Zhang, Zhiwu Qing, Mingqian Tang, Zhengrong Zuo, Changxin Gao, Rong Jin, and Nong Sang. Hybrid relation guided set matching for few-shot action recog- nition. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 19916–19925, 2022. ...

    2022

  39. [47]

    Few-shot learning meets transformer: Unified query-support transform- ers for few-shot classification

    Xixi Wang, Xiao Wang, Bo Jiang, and Bin Luo. Few-shot learning meets transformer: Unified query-support transform- ers for few-shot classification. IEEE Transactions on Circuits and Systems for Video Technology, 33(12):7789–7802, 2023

    2023

  40. [48]

    Task- aware dual-representation network for few-shot action recog- 10 nition

    Xiao Wang, Weirong Ye, Zhongang Qi, Guangge Wang, Jian- ping Wu, Ying Shan, Xiaohu Qie, and Hanzi Wang. Task- aware dual-representation network for few-shot action recog- 10 nition. IEEE Transactions on Circuits and Systems for Video Technology, 33(10):5932–5946, 2023

    2023

  41. [49]

    MoLo: Motion- augmented long-short contrastive learning for few-shot action recognition

    Xiang Wang, Shiwei Zhang, Zhiwu Qing, Changxin Gao, Yingya Zhang, Deli Zhao, and Nong Sang. MoLo: Motion- augmented long-short contrastive learning for few-shot action recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pag...

    2023

  42. [50]

    Cross-domain few-shot ac- tion recognition with unlabeled videos

    Xiang Wang, Shiwei Zhang, Zhiwu Qing, Yiliang Lv, Changxin Gao, and Nong Sang. Cross-domain few-shot ac- tion recognition with unlabeled videos. Computer Vision and Image Understanding (CVIU), 233:103737, 2023. 1, 3, 5, 7

    2023

  43. [51]

    Few-shot ac- tion recognition with captioning foundation models

    Xiang Wang, Shiwei Zhang, Hangjie Yuan, Yingya Zhang, Changxin Gao, Deli Zhao, and Nong Sang. Few-shot ac- tion recognition with captioning foundation models. arXiv preprint arXiv:2310.10125, 2023. 1, 4

    2023 arXiv

  44. [52]

    Few-shot action recognition via multi-view represen- tation learning

    Xiao Wang, Yang Lu, Wanchuan Yu, Yanwei Pang, and Hanzi Wang. Few-shot action recognition via multi-view represen- tation learning. IEEE Transactions on Circuits and Systems for Video Technology, pages 1–1, 2024. 2

    2024

  45. [53]

    CLIP-guided prototype modulating for few-shot action recognition

    Xiang Wang, Shiwei Zhang, Jun Cen, Changxin Gao, Yingya Zhang, Deli Zhao, and Nong Sang. CLIP-guided prototype modulating for few-shot action recognition. International Journal of Computer Vision, 132(6):1899–1912, 2024. 1, 3, 4

    1912

  46. [54]

    InternVideo: General video foundation models via generative and discriminative learning

    Yi Wang, Kunchang Li, Yizhuo Li, Yinan He, Bingkun Huang, Zhiyu Zhao, Hongjie Zhang, Jilan Xu, Yi Liu, Zun Wang, Sen Xing, Guo Chen, Junting Pan, Jiashuo Yu, Yali Wang, Limin Wang, and Yu Qiao. InternVideo: General video foundation models via generative and discriminative lear...

    2022 arXiv

  47. [55]

    Active exploration of multimodal complemen- tarity for few-shot action recognition

    Yuyang Wanyan, Xiaoshan Yang, Chaofan Chen, and Chang- sheng Xu. Active exploration of multimodal complemen- tarity for few-shot action recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 6492–6502, 2023. 1

    2023

  48. [56]

    Boosting few-shot action recognition with graph- guided hybrid matching

    Jiazheng Xing, Mengmeng Wang, Yudi Ruan, Bofan Chen, Yaowei Guo, Boyu Mu, Guang Dai, Jingdong Wang, and Yong Liu. Boosting few-shot action recognition with graph- guided hybrid matching. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 1...

    2023

  49. [57]

    Multiview Trans- formers for video recognition

    Shen Yan, Xuehan Xiong, Anurag Arnab, Zhichao Lu, Mi Zhang, Chen Sun, and Cordelia Schmid. Multiview Trans- formers for video recognition. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3323–3333, 2022. 2

    2022

  50. [58]

    Learning implicit temporal alignment for few-shot video classification

    Songyang Zhang, Jiale Zhou, and Xuming He. Learning implicit temporal alignment for few-shot video classification. International Joint Conference on Artificial Intelligence (IJ- CAI), pages 1309–1315, 2021. 6

    2021

  51. [59]

    Image BERT pre-training with online tokenizer

    Jinghao Zhou, Chen Wei, Huiyu Wang, Wei Shen, Cihang Xie, Alan Yuille, and Tao Kong. Image BERT pre-training with online tokenizer. In International Conference on Learning Representations (ICLR), 2021. 3

    2021

  52. [60]

    Compound memory networks for few-shot video classification

    Linchao Zhu and Yi Yang. Compound memory networks for few-shot video classification. In Proceedings of the European Conference on Computer Vision (ECCV) , pages 751–766,

  53. [62]

    Effect of Hyper-parameters on HTTN Number L and Parameters γ & β in TAA block. In Tab. S5 and Tab. S6, we explore the optimal TAA trans- former block number L and the parameter sharing strategy for parameters γ and β. Initially, L varies from 0 to 4, among which an L value of ...

  54. [63]

    Playing cards

    Generalization Verification To further validate the generalization of TAMT, we first con- duct experiments under a setting of generalization across datasets. Furthermore, we validate the effect of our TAMT L SSV2 Diving UCF Average Memory 0 53.41 42.87 94.97 63.58 1.2G 1 57.48...

  55. [64]

    Hammer phone

    Visualization Analyses To further validate the effectiveness of our TAMT method for addressing the problem of domain gap, we visualize feature heatmaps (the last layer of the backbone) of different models pre-trained on the source dataset (K-400) and those after tuning on the ...

  56. [2018]

    2, 6 1 TAMT: Temporal-Aware Model Tuning for Cross-Domain Few-Shot Action Recognition Supplementary Material In the supplementary materials, we first explore the effect of hyper-parameters on the Hierarchical Temporal Tuning Network (HTTN), mainly including the number L and pa...

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.