Pith. sign in

REVIEW 3 major objections 6 minor 29 references

DATTA: Domain-Adversarial Test-Time Adaptation for Cross-Domain WiFi-Based Human Activity Recognition

T0 review · 3 major / 6 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 The paper claims that its DATTA framework, combining domain-adversarial training, test-time feature-statistic alignment, and random weight resetting, lifts cross-domain WiFi-based human activity recognition to 68.13% F1 on Widar3.0-G6D…

desk verdict Solid incremental WiFi HAR paper with thorough ablations and code; the 8.1% SotA claim is inflated by asymmetric tuning, but the core DATTA combination is a real, useful contribution. read the letter →

arxiv 2411.13284 v1 pith:AKKSHZ5C submitted 2024-11-20 cs.CV cs.AIcs.ETcs.LG

Julian Strohmayer , Rafael Sterzinger , Matthias Wödlinger , Martin Kampel This is my paper · ORCID
classification cs.CVcs.AIcs.ETcs.LG
keywords WiFisensingChannelStateInformationHumanActivityRecognitiondomain-adversarialtrainingtest-timeadaptationrandomweightresettingcross-domaingeneralization
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 claims that a framework called DATTA, which combines domain-adversarial training (DAT), test-time adaptation (TTA), and random weight resetting, lets a WiFi-based human activity recognition model keep working when deployed in an unseen environment, with unfamiliar users, or on different hardware. On the Widar3.0-G6D benchmark, DATTA reaches an F1-Score of 68.13%, beating a naive combination of a state-of-the-art video TTA method and WiFi DAT by 8.1 percentage points. The model is lightweight, with about 40k parameters, and fast enough for online adaptation, so the authors argue the approach is practical for real-time use. If correct, this would mean the three components are complementary: DAT learns domain-invariant features, TTA aligns statistics of a new domain at inference, and weight resetting keeps the model from forgetting what it learned.

What carries the argument

The mechanism combines a gradient reversal layer from domain-adversarial training with a test-time loss that matches per-layer feature-map statistics to source statistics. In DAT, a domain discriminator is attached to the class-token embedding, and the gradient reversal layer reverses its gradients so the feature extractor learns representations the discriminator cannot tell apart across domains. At test time, DATTA updates the model by minimizing $L_{\text{TTA}} = \sum_{l\in L} \|\mu_l(B;\theta) - \bar\mu_l\|^2 + \|\sigma^2_l(B;\theta) - \bar\sigma^2_l\|^2$, which aligns the first transformer encoder layer's feature-map statistics with the source statistics, using exponential moving averages for online batches of size one. Random weight resetting then restores a Bernoulli-sampled subset of weights to their source values, with a reset rate of $p = 1 \times 10^{-4}$, preventing drift. The feature extractor is a WiFlexFormer with a convolutional stem, Gaussian positional encoding, and a four-layer transformer encoder, chosen for its roughly 40k parameter count.

What would settle it

Run DATTA on a cross-domain split in which the target domain differs strongly in hardware or room geometry from the source, and compare aligning the first transformer layer against aligning each of the other three layers and against using batch-norm running statistics: if the first-layer anchor does not consistently outperform the alternatives and sometimes degrades performance, the paper's central adaptation mechanism is not doing the claimed work.

Watch

Extended reading notes

Core claim

The central claim is that DATTA's three components cooperate: domain-adversarial training makes the learned features invariant across source domains, test-time adaptation aligns the first transformer encoder layer's feature-map mean and variance to source statistics, and random weight resetting continually pulls a fraction of parameters back to the source model to prevent catastrophic forgetting. In the paper's evaluation, removing any component hurts, and only the full WDATTA model with augmentation and weight resetting achieves the highest F1 of 68.13%. The authors show that weight resetting helps only when the model is already domain-invariant, and that TTA alone without DAT gives little benefit. They also report that the method recovers quickly after a difficult domain and stays stable under prolonged adaptation to one domain.

Load-bearing premise

The load-bearing assumption is that matching the first transformer encoder layer's feature-map mean and variance to source statistics is always a safe and sufficient adaptation signal for CSI, so that the TTA updates help rather than hurt under large or rapid domain shifts.

Editorial extensions

If this is right

  • WiFi-based human activity recognition can adapt online to new rooms, subjects, and hardware without labeled target data, using only unlabeled CSI streams.
  • Cross-domain performance on Widar3.0-G6D improves to 68.13% F1, which is 8.1% higher than a naive combination of existing DAT and video-derived TTA methods.
  • Random weight resetting is useful only after the model already has domain-invariant features; on a purely source-trained model it can undo helpful adaptation.
  • The full DATTA pipeline runs at about 20 ms per sample on an RTX 2070 and about 115 ms on a Jetson Orin Nano, keeping real-time human activity recognition feasible.

Reading between the lines

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

  • The paper leaves implicit that aligning only the first transformer encoder layer hints at a property of CSI spectrograms: domain-specific variation may concentrate in the earliest feature maps, so testing deeper layers as anchors would clarify whether this is a general principle or a quirk of this benchmark.
  • The 8.1% gap over a naive ViTTA-plus-DAT combination suggests that porting video TTA losses to WiFi requires reshaping the loss and target initialization, not just changing the input modality; CSI-specific augmentation and loss design may matter more than the raw alignment objective.
  • A testable extension would be to apply DATTA to other RF sensing modalities, such as radar or mmWave, where the same channel-statistics shift problem arises, or to trigger weight resetting only when a domain-shift detector signals drift.
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 / 6 minor

Summary. The paper proposes DATTA, a framework for cross-domain WiFi-based human activity recognition that combines domain-adversarial training (DAT), test-time adaptation (TTA), and random weight resetting on a WiFlexFormer backbone. DATTA aligns feature-map statistics at test time with source statistics (Eq. 10), updating parameters online with an exponential moving average (Eqs. 11–12), and periodically resets a random subset of weights to source values (Eq. 13) to mitigate catastrophic forgetting. The authors evaluate on a modified subset of Widar3.0, called Widar3.0-G6D, with disjoint training and test domains, and report F1-Score improvements from each component: augmentation, the confidence control constraint, and weight resetting. They also report inference times on GPU and edge hardware. The headline result is an "8.1% higher F1-Score" over a naive combination of a video-based TTA method and a WiFi DAT method.

Significance. If the results hold, DATTA is a practical contribution: it demonstrates that combining offline domain-adversarial training with online test-time feature-statistic alignment is feasible for low-parameter WiFi sensing models, and the public code and dataset-split script support reproducibility. The paper includes a component-wise ablation, reports means and standard deviations over three runs, and measures real-time latency on an edge device, which are useful strengths. However, the central external claim of SotA superiority is weakened by an asymmetric tuning protocol and a single-dataset evaluation, so the significance of the claimed 8.1% improvement is not yet established.

major comments (3)
  1. [§4.3, Table 4 and Abstract] The headline "8.1% higher F1-Score" is a relative improvement (68.13 vs. 63.04, i.e., 5.09 percentage points) and is not a fair SotA comparison as configured. WDATTA receives Bayesian and grid hyperparameter tuning (Section 4.2), while the W+both baseline and the SotA component variants are only described as "slightly tuned" (Table 4 note). Moreover, W+DAT alone achieves F1 63.16, slightly higher than W+both at 63.04, so the chosen baseline is arguably degraded by the added ViTTA component. The comparison therefore confounds the method with tuning effort and baseline integration, and the abstract's claim of an 8.1% improvement over SotA is unsupported as stated. The authors should retune the baselines with the same search budget and compare against the strongest component combination, or explicitly recast the 8.1% as a relative gain over their own naive baseline rather than over SotA.
  2. [§4.1 and §4.3] The evaluation rests on a single curated dataset, Widar3.0-G6D, with one room-participant split. The test set contains only one environment and is shuffled, which the authors themselves describe as an extreme high-frequency shift case. The claimed cross-domain generalizability and the "comprehensive evaluation" framing would be substantially stronger with a second dataset or at least a second split/protocol. In addition, the first continuous-domain experiment shows a significant drop on domain D8, which the authors attribute to likely mislabeled data without confirmation; this acknowledged limitation means the reported averages may be sensitive to label noise in one domain. These issues do not invalidate the internal ablations, but they do limit the external validity of the central SotA claim.
  3. [§3.2] The choice of aligning only the first transformer encoder layer is stated as an observation ("we observe that optimal results are obtained") but is not supported by a reported ablation or sensitivity analysis in the main text. Since the effectiveness of TTA depends on this anchor, and the paper elsewhere motivates the method as a general framework, the missing layer-selection experiment is a load-bearing gap. The authors should report a layer ablation (e.g., first vs. middle vs. last layers, and combinations) and, if space permits, a sensitivity analysis for the reset rate p, to show that the chosen configuration is not a fragile point in hyperparameter space.
minor comments (6)
  1. [Abstract and §4.3] The phrase "8.1% higher F1-Score" is ambiguous; since F1-Score is already a percentage, the authors should state explicitly that this is a relative improvement over the 63.04 F1-Score of W+both, with an absolute gain of 5.09 percentage points.
  2. [§3.2, Eq. (8)] Equation (8) uses x inside the expectation over φ_l(x; θ) while the surrounding text defines inputs as spectrograms s; the notation should be made consistent.
  3. [Table 4 note] The phrase "slightly tuned hyperparameters" is not specific enough for a SotA comparison; the authors should list which hyperparameters were tuned, their values, and the search budget used for each baseline so that the comparison is reproducible.
  4. [§4.3, Figure 3] The rolling-window F1 curves in Figure 3 are shown without error bars or confidence intervals; given that the aggregate results report standard deviations over three runs, adding uncertainty bands or at least stating that the curves are from a single representative run would improve clarity.
  5. [§4.3, Table 5] The claim that 115.39 ms per sample on the Jetson Orin Nano is "sufficient for HAR" should be justified against a concrete latency or throughput requirement for the intended real-time application, since 9 frames per second is marginal for some activity recognition scenarios.
  6. [§5] The conclusion states "we evaluated a component of DATTA" where the intended wording appears to be "each component"; this typo should be corrected.

Circularity Check

0 steps flagged · score 1.0 of 10

Empirical composition paper; no load-bearing circularity, though the SotA claim is weakened by uneven tuning effort.

full rationale

The paper is an empirical evaluation and combination of existing components (DAT from [11], TTA from [12], weight resetting from [27]), not a derivation in which a predicted quantity is defined as its input. The abstract's "8.1% higher F1-Score" is a measured comparison on the held-out Test split of Widar3.0-G6D (Table 4: WDATTA 68.13 vs W+both 63.04); the quantity reported is F1 on activity labels, not the TTA objective itself, so the comparison does not reduce to Eq. 10 by construction. TTA hyperparameters are tuned on ValTTA, a disjoint domain split, and Test is shuffled as an extreme shift scenario, so the headline result is not a fitted-input-called-prediction in the strict sense. The paper does cite the authors' own prior work for the WiFlexFormer backbone [23], the augmentation module [22], and an earlier ESP32 sensing study [21], but none of these citations is used to justify the central comparison: the ablation tables report measured outcomes for augmentations, CCC, and weight resetting, and the external DAT and ViTTA baselines are evaluated independently of those self-citations. There is a legitimate evaluation-design concern, flagged by the skeptic pass: WDATTA is tuned with Bayesian and grid search while the SotA baselines are only "slightly tuned" (Table 4), so the claimed 8.1% gain may partly reflect tuning effort rather than the framework itself. That is a correctness/interpretation risk, not circularity, because the tested target domains are disjoint from the tuning split and the reported F1 values are genuine held-out measurements. No uniqueness theorem is imported from the authors, no ansatz is smuggled in via self-citation, and no known result is merely renamed. Overall, the derivation chain is self-contained enough that the central claim, however debatable in experimental strength, is not circular.

Assumptions & free parameters 8 free parameters · 4 assumptions · 0 invented entities

The core claim depends on a small set of hyperparameters and standard domain-adaptation assumptions. All numerical hyperparameters were tuned on validation data, which is legitimate but means the reported gains are conditional on that tuning. No new physical entities are introduced.

free parameters (8)
  • alpha (DAT loss weight) = 0.3
    Weight for domain adversarial loss in Eq. 7; selected via hyperparameter search.
  • beta (CCC loss weight) = 0.2
    Weight for confidence control constraint in Eq. 7; chosen by search and ablated in Table 3.
  • gamma (GRL scaling) = 8
    Scaling in lambda schedule (Eq. 5) controlling adversarial signal strength.
  • weight reset rate p = 1e-4
    Bernoulli probability for resetting parameters to source values during TTA (Eq. 13); only this rate was evaluated.
  • TTA alignment layer set L = first of four transformer encoder layers
    Selected experimentally as the best layer for statistic alignment in Eq. 10.
  • TTA learning rate = 1e-6 (SotA comparison; not stated for DATTA)
    Learning rate for TTA updates; specified only for the ViTTA comparison, not for DATTA in the main text.
  • EMA momentum in Eqs. 11-12 = not specified
    Momentum parameter for online mean/variance estimates; value not given in the paper.
  • Augmentation hyperparameters = not specified
    Amplitude perturbation scale, rotation range, and dropout rates in the augmentation module are not detailed in the main text.
assumptions (4)
  • domain assumption Domain labels are available during training and room-participant combinations define meaningful domain identities.
    Section 3.1 uses domain labels d in Eq. 2 for the discriminator; the dataset split constructs domains from room-participant pairs.
  • domain assumption Source feature statistics (mean and variance) computed on the training set are a valid alignment target at test time.
    The TTA loss in Eq. 10 aligns online statistics to source statistics; this is the core transfer assumption from video TTA [12].
  • ad hoc to paper The first transformer encoder layer's feature maps are the most informative for domain alignment.
    The paper states in Section 3.2 that 'optimal results are obtained by selecting L to contain only the first out of four transformer encoder layers'; this is a tuned choice, not derived.
  • ad hoc to paper Random weight resetting with a small probability prevents catastrophic forgetting without destroying useful test-time adaptation.
    Eq. 13 and the discussion in Section 4.3 propose this heuristic; no theoretical guarantee is given, only empirical support.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DATTA: Domain-Adversarial Test-Time Adaptation for Cross-Domain WiFi-Based Human Activity Recognition." pith.science (2026). https://pith.science/paper/AKKSHZ5C

@misc{pith2026241113284,
  author       = {Pith},
  title        = {Pith review of: DATTA: Domain-Adversarial Test-Time Adaptation for Cross-Domain WiFi-Based Human Activity Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AKKSHZ5C}},
  note         = {Machine review of arXiv:2411.13284}
}
read the original abstract

Cross-domain generalization is an open problem in WiFi-based sensing due to variations in environments, devices, and subjects, causing domain shifts in channel state information. To address this, we propose Domain-Adversarial Test-Time Adaptation (DATTA), a novel framework combining domain-adversarial training (DAT), test-time adaptation (TTA), and weight resetting to facilitate adaptation to unseen target domains and to prevent catastrophic forgetting. DATTA is integrated into a lightweight, flexible architecture optimized for speed. We conduct a comprehensive evaluation of DATTA, including an ablation study on all key components using publicly available data, and verify its suitability for real-time applications such as human activity recognition. When combining a SotA video-based variant of TTA with WiFi-based DAT and comparing it to DATTA, our method achieves an 8.1% higher F1-Score. The PyTorch implementation of DATTA is publicly available at: https://github.com/StrohmayerJ/DATTA.

Figures

Figures reproduced from arXiv: 2411.13284 by the authors.

Figure 1
Figure 1. Overview of the domain-adversarial training approach used in DATTA. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Widar3.0 recording setup featuring a single transmit [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. TTA performance across continuous test domain se [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 21 canonical work pages

  1. [1]

    Cross-domain wifi sensing with channel state information: A survey

    Chen Chen, Gang Zhou, and Youfang Lin. Cross-domain wifi sensing with channel state information: A survey. ACM Computing Surveys, 55(11):1–37, 2023. 1

    work page 2023

  2. [2]

    Adversarial Continual Learn- ing

    Sayna Ebrahimi, Franziska Meier, Roberto Calandra, Trevor Darrell, and Marcus Rohrbach. Adversarial Continual Learn- ing. In Computer Vision – ECCV 2020 , pages 386–402. Springer International Publishing, Cham, 2020. Series Ti- tle: Lecture Notes in Computer Science. 3, 4

    work page 2020

  3. [3]

    Sensing technology for human activity recognition: A comprehensive survey

    Biying Fu, Naser Damer, Florian Kirchbuchner, and Arjan Kuijper. Sensing technology for human activity recognition: A comprehensive survey. IEEE Access, PP:1–1, 2020. 1

    work page 2020

  4. [4]

    Test-Time Training with Masked Autoencoders

    Yossi Gandelsman, Yu Sun, Xinlei Chen, and Alexei Efros. Test-Time Training with Masked Autoencoders. InAdvances in Neural Information Processing Systems , pages 29374– 29385. Curran Associates, Inc., 2022. 5

    work page 2022

  5. [5]

    Yossi Gandelsman, Yu Sun, Xinlei Chen, and Alexei A. Efros. Test-time training with masked autoencoders, 2022. 2

    work page 2022

  6. [6]

    Unsupervised domain adaptation by backpropagation

    Yaroslav Ganin and Victor Lempitsky. Unsupervised domain adaptation by backpropagation. In International conference on machine learning, pages 1180–1189. PMLR, 2015. 2

    2015

  7. [7]

    Domain-adversarial training of neural networks

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pas- cal Germain, Hugo Larochelle, Franc ¸ois Laviolette, Mario March, and Victor Lempitsky. Domain-adversarial training of neural networks. Journal of machine learning research , 17(59):1–35, 2016. 2

    2016

  8. [8]

    Tool release: Gathering 802.11n traces with chan- nel state information

    Daniel Halperin, Wenjun Hu, Anmol Sheth, and David Wetherall. Tool release: Gathering 802.11n traces with chan- nel state information. Computer Communication Review, 41: 53, 2011. 5

    work page 2011

Show all 29 references
  1. [9]

    Hernandez and Eyuphan Bulut

    Steven M. Hernandez and Eyuphan Bulut. Wifi sensing on the edge: Signal processing techniques and challenges for real-world systems. IEEE Communications Surveys & Tuto- rials, 25(1):46–76, 2023. 1

    2023

  2. [10]

    Rfboost: Understanding and boosting deep wifi sensing via physical data augmentation

    Weiying Hou and Chenshu Wu. Rfboost: Understanding and boosting deep wifi sensing via physical data augmentation. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 8:1 – 26, 2024. 5

    2024

  3. [11]

    Towards environment inde- pendent device free human activity recognition

    Wenjun Jiang, Chenglin Miao, Fenglong Ma, Shuochao Yao, Yaqing Wang, Ye Yuan, Hongfei Xue, Chen Song, Xin Ma, Dimitrios Koutsonikolas, et al. Towards environment inde- pendent device free human activity recognition. In Proceed- ings of the 24th annual international conference ...

    2018

  4. [12]

    Video Test-Time Adaptation for Action Recognition

    Wei Lin, Muhammad Jehanzeb Mirza, Mateusz Kozinski, Horst Possegger, Hilde Kuehne, and Horst Bischof. Video Test-Time Adaptation for Action Recognition. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 22952–22961, Vancouver, BC, Canada, 202...

    2023

  5. [13]

    Human activ- ity sensing with wireless signals: A survey

    Jiao Liu, Guanlong Teng, and Feng Hong. Human activ- ity sensing with wireless signals: A survey. Sensors, 20(4),

  6. [14]

    Ttt++: When does self-supervised test-time training fail or thrive? Advances in Neural Information Processing Systems , 34: 21808–21820, 2021

    Yuejiang Liu, Parth Kothari, Bastien Van Delft, Baptiste Bellot-Gurlet, Taylor Mordan, and Alexandre Alahi. Ttt++: When does self-supervised test-time training fail or thrive? Advances in Neural Information Processing Systems , 34: 21808–21820, 2021. 2

    2021

  7. [15]

    Conditional adversarial domain adapta- tion

    Mingsheng Long, Zhangjie Cao, Jianmin Wang, and Michael I Jordan. Conditional adversarial domain adapta- tion. Advances in neural information processing systems , 31, 2018. 2

    2018

  8. [16]

    Improved self-training for test-time adaptation

    Jing Ma. Improved self-training for test-time adaptation. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 23701–23710, 2024. 2, 5

    2024

  9. [17]

    Wifi sens- ing with channel state information: A survey

    Yongsen Ma, Gang Zhou, and Shuangquan Wang. Wifi sens- ing with channel state information: A survey. ACM Comput. Surv., 52(3), 2019. 1

    2019

  10. [18]

    Jehanzeb Mirza, Jakub Micorek, Horst Possegger, and Horst Bischof

    M. Jehanzeb Mirza, Jakub Micorek, Horst Possegger, and Horst Bischof. The Norm Must Go On: Dynamic Unsu- pervised Domain Adaptation by Normalization. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 14745–14755, New Orleans, LA, USA, 2022. IEEE. 5

    2022

  11. [19]

    Moment matching for multi-source domain adaptation

    Xingchao Peng, Qinxun Bai, Xide Xia, Zijun Huang, Kate Saenko, and Bo Wang. Moment matching for multi-source domain adaptation. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 1406–1415,

  12. [20]

    Omer Gokalp Serbetci, Ju-Hyung Lee, Daoud Burghal, and Andreas F. Molisch. Simple and effective augmentation methods for csi based indoor localization, 2023. 1

    2023

  13. [21]

    Wifi csi- based long-range through-wall human activity recognition with the esp32

    Julian Strohmayer and Martin Kampel. Wifi csi- based long-range through-wall human activity recognition with the esp32. In Computer Vision Systems, pages 41–50, Cham, 2023. Springer Nature Switzerland. 1

    2023

  14. [22]

    Wifi CSI-based long- range person localization using directional antennas

    Julian Strohmayer and Martin Kampel. Wifi CSI-based long- range person localization using directional antennas. In The Second Tiny Papers Track at ICLR 2024, 2024. 3

    2024

  15. [23]

    Wiflexformer: Efficient wifi-based person-centric sens- ing, 2024

    Julian Strohmayer, Matthias W ¨odlinger, and Martin Kam- pel. Wiflexformer: Efficient wifi-based person-centric sens- ing, 2024. 3

    2024

  16. [24]

    Test-time training with self- supervision for generalization under distribution shifts

    Yu Sun, Xiaolong Wang, Zhuang Liu, John Miller, Alexei Efros, and Moritz Hardt. Test-time training with self- supervision for generalization under distribution shifts. In International conference on machine learning, pages 9229–

  17. [25]

    Learning to (Learn at Test Time): RNNs with Expressive Hidden States, 2024

    Yu Sun, Xinhao Li, Karan Dalal, Jiarui Xu, Arjun Vikram, Genghan Zhang, Yann Dubois, Xinlei Chen, Xiaolong Wang, Sanmi Koyejo, Tatsunori Hashimoto, and Carlos Guestrin. Learning to (Learn at Test Time): RNNs with Expressive Hidden States, 2024. arXiv:2407.04620. 2, 5

    2024 arXiv

  18. [26]

    Adversarial discriminative domain adaptation

    Eric Tzeng, Judy Hoffman, Kate Saenko, and Trevor Darrell. Adversarial discriminative domain adaptation. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 7167–7176, 2017. 2

    2017

  19. [27]

    Con- tinual Test-Time Domain Adaptation

    Qin Wang, Olga Fink, Luc Van Gool, and Dengxin Dai. Con- tinual Test-Time Domain Adaptation. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 7191–7201, New Orleans, LA, USA, 2022. IEEE. 2, 4 9

    2022

  20. [28]

    Efros, and Xiaolong Wang

    Renhao Wang, Yu Sun, Yossi Gandelsman, Xinlei Chen, Alexei A. Efros, and Xiaolong Wang. Test-time training on video streams, 2023. 2

    2023

  21. [29]

    Widar3.0: Zero-effort cross-domain gesture recognition with wi-fi

    Yi Zhang, Yue Zheng, Kun Qian, Guidong Zhang, Yunhao Liu, Chenshu Wu, and Zheng Yang. Widar3.0: Zero-effort cross-domain gesture recognition with wi-fi. IEEE Transac- tions on Pattern Analysis and Machine Intelligence, 44(11): 8671–8688, 2022. 1, 5 10

    2022

Pith tools

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