Pith. sign in

REVIEW 5 major objections 6 minor 50 references

Unsupervised Domain Adaptation with Dynamic Clustering and Contrastive Refinement for Gait Recognition

T0 review · 5 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read GaitDCCR establishes that unsupervised gait recognition can surpass fully supervised methods, reporting 87.4% mean rank-1 on CASIA-B.

desk verdict Strong empirical pipeline for unsupervised gait recognition, but the unstated cluster count C undermines the fully-unsupervised framing until it is disclosed. read the letter →

arxiv 2501.16608 v2 pith:GVA3JCMY submitted 2025-01-28 cs.CV

Xiaolei Liu , Yan Sun , Zhiliang Wang , Mark Nixon This is my paper · ORCID
classification cs.CV
keywords gaitrecognitionunsuperviseddomainadaptationpseudo-labelnoisedynamicclusteringcontrastivelearningteacher-studenttrainingsilhouettedataaugmentationsoftpseudo-labels
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

This paper tries to establish that unsupervised gait recognition can become nearly as accurate as supervised recognition if the noise in clustering-generated pseudo-labels is treated explicitly. It proposes GaitDCCR, which combines dynamic clustering thresholds, density-weighted cluster centroids, confidence-based soft pseudo-labels, and a contrastive teacher branch with silhouette augmentation. On CASIA-B with a GaitGL backbone it reports 87.4% mean rank-1, 8.2 points above the prior unsupervised state of the art, and above some fully supervised baselines. If this holds, label-free gait identification from silhouette sequences becomes practical for surveillance and security applications.

What carries the argument

The method's central object is a soft pseudo-label pipeline built on a memory bank of cluster centroids. In the clustering stage, an exponential decay rule $\epsilon = \epsilon_0 \eta^{\text{epochs}}$ adjusts the cluster acceptance threshold, and each centroid is recomputed as a density-weighted sum $m'_i = \sum_{x_i\in C_i} w_i f_i$, where $w_i$ weights central samples more than edge samples. In training, a confidence matrix $F_{i,j} = p_{i,j}/\sum_j p_{i,j}$ with $p_{i,j}=\sigma(-d(i,j))$ converts hard labels into soft labels $\hat{y}_i = \alpha y_i + (1-\alpha)F_i$, and a teacher branch, updated by exponential moving average, applies dilation and erosion augmentation to the body region, finds the $k$ nearest centroids, and produces the final soft label $y_i^s = \beta P_i + (1-\beta)\hat{y}_i$; the student is trained with an InfoNCE loss against the memory bank. This machinery carries the argument because every reported gain is attributed to increasingly reliable pseudo-labels and centroids.

What would settle it

Run GaitDCCR on a target set whose true identity count is withheld, estimating C from the feature distribution by silhouette analysis or a density-based cluster count, and report rank-1; if accuracy drops sharply relative to the run that sets C to the known training-split identity count, the central claim of fully label-free adaptation is falsified.

Watch

Extended reading notes

Core claim

The central claim is that pseudo-label noise caused by cross-domain clothing and viewpoint changes is the dominant failure mode in unsupervised gait recognition, and that it can be suppressed by a four-part mechanism: an exponentially decaying clustering threshold, density-weighted centroids, confidence-based refinement of one-hot pseudo-labels into soft labels, and a teacher-student contrastive module that uses body-region silhouette augmentation to find the true clusters of clothing-augmented samples. The paper reports that this mechanism lifts GaitGL-based unsupervised recognition on CASIA-B to 87.4% mean rank-1, surpassing prior unsupervised methods by 8.2% and some fully supervised methods, with the largest gains in the clothing-change (CL) condition, 74.2%. On OUMVLP, the method reaches 68.5% mean rank-1 when GREW is the source dataset.

Load-bearing premise

The load-bearing premise is that the number of clusters C is fixed correctly before training; the paper never states how C is chosen, and since the target dataset is unlabeled, a deployment would have to guess the number of identities, so the reported 87.4% may depend on knowing that number in advance.

Editorial extensions

If this is right

  • On CASIA-B with a GaitGL backbone, GaitDCCR reaches 87.4% mean rank-1, outperforming UGRSF, STANet, and GOUDA, and beating some fully supervised methods, especially in the CL clothing condition.
  • On OUMVLP as target, the method reports 68.5% mean rank-1 with a GaitSet backbone and 60.6% with GaitGL when GREW is the source, a jump of over 30 points over GOUDA.
  • Ablation results attribute separate gains to each component: dynamic clustering parameters and dynamic weighted centroids improve clustering quality, while confidence-based pseudo-label refinement and the contrastive teacher module contribute the larger share of the accuracy increase.
  • The exponential decay schedule for the clustering threshold beats fixed thresholds and other decay curves, indicating that adapting the clustering acceptance criterion during training is a direct source of improvement.
  • The method works with both GaitSet and GaitGL backbones, suggesting the refinement and dynamic clustering strategies are backbone-agnostic rather than tied to one architecture.

Reading between the lines

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

  • The reported results likely rely on knowing the number of identities in the target training split, 74 on CASIA-B and 5,153 on OUMVLP; on a genuinely unlabeled deployment this number must be estimated, so the practical gain may be smaller than the claimed 8.2 points.
  • The same dynamic-threshold and confidence-refinement recipe should transfer to person re-identification and other fine-grained biometrics, where clustering-based pseudo-label noise is also the shared bottleneck.
  • The large source-dataset gap on OUMVLP, 49.2% with a CASIA-B source versus 68.5% with a GREW source, suggests that source diversity, not only label refinement, sets the ceiling; a natural extension is self-supervised pretraining on unlabeled in-the-wild gait video.
  • A testable prediction is that replacing the teacher branch's body-only augmentation with augmentation that also perturbs head and leg regions would lower CL-condition accuracy, because those regions are nearly unaffected by real clothing changes.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The manuscript proposes GaitDCCR, an unsupervised domain adaptation method for gait recognition. The method has two stages: a clustering stage with Dynamic Clustering Parameters (DCP) and Dynamic Weighted Centroids (DWC), and a training stage with Confidence-based Pseudo-label Refinement (CPR) and a Contrastive Teacher Module (CTM). The authors evaluate on CASIA-B and OUMVLP using GaitSet and GaitGL backbones, reporting strong rank-1 accuracies, including 87.4% mean rank-1 on CASIA-B with GaitGL, which they claim surpasses prior state-of-the-art unsupervised methods and some fully supervised methods.

Significance. If the reported results hold, GaitDCCR would be a substantial empirical advance for unsupervised gait recognition, particularly for the challenging clothing-change scenario. The paper includes component-level ablations, an analysis of dynamic thresholds, and promises code release, which are strengths. However, the central claim is weakened by the undisclosed cluster count C, an internal contradiction between the text and Table 4, and an inconsistency between the described InfoNCE loss and the actual cross-entropy formulation. These issues need to be resolved before the results can be considered reproducible or the unsupervised claim fully verified.

major comments (5)
  1. [Section 3, Section 4.2, Algorithm 1] The number of clusters C is never defined or reported anywhere, yet it is load-bearing for the entire pipeline: pseudo-labels are defined as yi ∈ {1,...,C}, the memory bank contains C centroids, F has dimensions R^{B×C} in Eq. (8), Eq. (10) assigns probabilities over C clusters, and Eq. (13) sums the loss over C clusters. The paper must state how C is chosen. If C is set to the number of known identities in the target training split (74 for CASIA-B and 5,153 for OUMVLP), that relies on information a genuinely unlabeled deployment setting would not have; this would need to be acknowledged and ablated. If C is estimated by another mechanism, that mechanism must be described. The dynamic epsilon schedule in Eq. (4) only changes cluster tightness and cannot correct an incorrect cluster count, so the omission is not benign.
  2. [Section 5.2.1 and Table 4] The text states that "using the GaitGL backbone consistently yields better accuracy compared to GaitSet in all tested conditions." This is directly contradicted by Table 4 for GREW→OUMVLP, where GaitSet reaches 68.5% mean rank-1 and GaitGL reaches 60.6%. The claim should be corrected or qualified, and the discrepancy between the two backbones in the GREW source setting should be discussed.
  3. [Sections 3 and 4.6] The method is described as using InfoNCE loss, and Algorithm 1 line 18 says "Computing InfoNCE loss," but the actual loss in Eq. (13) is a cross-entropy loss lce(Mq fi, y^s_i) applied to soft pseudo-labels. Section 3 Eq. (2) defines a separate InfoNCE loss that is not used in the final training objective. The authors should clarify which loss is actually optimized and reconcile the notation, since the distinction matters for reproducibility.
  4. [Section 4.3, Eqs. (5)-(7)] The text says "samples with smaller density distance are more likely to be real samples, while samples with larger density distance are likely to be noise samples," but the defined quantity Di = σ(−(1/|Ci|) Σ d(i,j)) is larger when the average distance to other samples in the cluster is smaller. Thus, a "larger density distance" as defined corresponds to a sample closer to the cluster center, not a noisier sample. This makes the weighting direction in Eq. (6) ambiguous and should be corrected or clarified.
  5. [Section 5.3.2 and Table 5] The ablation for the contrastive teacher module with and without data augmentation is unclear. Table 5 has columns for DCP, DWC, CPR, CTM, and CTM (w/o DA), but rows 7 and 8 both show four checkmarks and no visible distinction between the configurations. The text refers to experiments "7" and "8" for the DA ablation, but a reader cannot tell which row uses DA and which does not. Please fix the table formatting and explicitly annotate each row.
minor comments (6)
  1. [Section 5.3.2] The subsection title says "Effectiveness of CPR and CTM in the clustering phase," but the components described and ablated are in the training phase; please correct the wording.
  2. [Section 4.5.2, Eq. (10)] Equation (10) has an unclear denominator, "d(˜xi,C j)P fCk i", which appears to be a formatting or notation error. Please rewrite the normalization explicitly.
  3. [Algorithm 1 and Eq. (11)] Algorithm 1 line 17 uses y^k_i in the soft-label update, whereas Eq. (11) defines ys_i = β·Pi + (1−β)·y^i. The notation should be made consistent.
  4. [General] The manuscript header on page 2 reads "Preprint submitted to Nuclear Physics B," which appears to be an incorrect journal name for this computer vision paper; please update it.
  5. [Table 1 caption] The caption in Table 1 says "CAISA-B" instead of "CASIA-B"; please fix the typo.
  6. [Section 5.2] Please cite the actual publication venues for GOUDA, UGRSF, and STANet in the comparison tables rather than only arXiv/venue names in the table cells, and double-check the reference formatting for these entries.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the paper reports benchmark evaluations of a new method; its self-citations are not load-bearing and the reported accuracies do not reduce to the method's own inputs.

full rationale

GaitDCCR is an empirical method paper. Its central claim—87.4% mean rank-1 on CASIA-B with GaitGL, 8.2% over SOTA—is supported by benchmark evaluations against external baselines and standard train/test splits; the result is not derived from fitted constants or from equations that tautologically produce it. The pipeline equations (4)–(13) define the method rather than encoding the reported accuracy. Self-citations such as [15] and [30] appear only as examples in related-work enumerations and are not load-bearing; no uniqueness theorem or prior-work ansatz is invoked to force the design. The omission of how cluster count C is chosen is a reproducibility and honesty concern for the 'unsupervised' claim, but the paper never states C = 74 or otherwise reduces the result to known split information, so it is not a demonstrated circularity. Hyperparameters (epsilon0, eta, alpha, beta, k) are selected on the target benchmarks, which is a standard practice that weakens external validity but is not structural circularity. Verdict: no significant circularity.

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

The central claim rests on several hand-chosen hyperparameters (epsilon0, eta, alpha, beta, k, gamma, mu) and three domain assumptions: density-distance separating noise, clothing augmentation approximating real clothing change, and the k-nearest-cluster strategy recovering true identities. The unstated cluster-count assumption is the most consequential because the whole pseudo-labeling and memory-bank machinery depends on C.

free parameters (8)
  • epsilon0 (initial clustering threshold) = 0.8
    Initial threshold for clustering pseudo-labels; chosen by hand, no sensitivity analysis beyond the decay schedule.
  • eta (clustering threshold decay rate) = 0.97
    Exponential decay rate for epsilon; selected to match the model learning curve on the benchmark.
  • alpha (CPR blending weight) = 0.4
    Weight between the one-hot pseudo-label and the confidence vector in Eq. 9; tuned on evaluation datasets.
  • beta (teacher-student blend weight) = 0.4
    Weight between teacher latent-cluster probabilities and refined pseudo-label in Eq. 11; tuned.
  • k (Latent Cluster Set size) = 2
    Number of nearest clusters considered in CTM; chosen by ablation on CASIA-B.
  • gamma (teacher EMA momentum) = 0.99
    Exponential moving average factor for teacher encoder parameters.
  • mu (memory bank momentum) = 0.2
    Momentum factor for updating cluster centroids in Eq. 3, inherited from the baseline.
  • tau (InfoNCE temperature) = not reported
    Temperature in Eq. 2 is never given; it directly scales the contrastive loss and affects training dynamics.
assumptions (4)
  • domain assumption Noisy samples tend to lie at cluster edges, and density distance separates them from reliable samples.
    Sec. 4.3 states 'Empirical analysis shows...' without providing the analysis; the DWC centroid weighting depends on this premise.
  • domain assumption Clothing variation can be approximated by dilation and erosion of the body region of binary silhouettes.
    Sec. 4.5.1 claims this simulates NM-to-CL transitions; the CTM teacher module depends on this approximation being faithful.
  • ad hoc to paper The k nearest clusters to an augmented sample contain the true identity of that sample.
    Sec. 4.5.2 builds the Latent Cluster Set from the k nearest clusters and uses it as a soft label; no independent evidence is given that k=2 clusters capture the true identity.
  • domain assumption The number of target-domain clusters C is known or correctly set.
    Sec. 3 defines pseudo-labels over C clusters and initializes the memory bank with C centroids, but no procedure is given for choosing C in a truly unlabeled target set.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unsupervised Domain Adaptation with Dynamic Clustering and Contrastive Refinement for Gait Recognition." pith.science (2026). https://pith.science/paper/GVA3JCMY

@misc{pith2026250116608,
  author       = {Pith},
  title        = {Pith review of: Unsupervised Domain Adaptation with Dynamic Clustering and Contrastive Refinement for Gait Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GVA3JCMY}},
  note         = {Machine review of arXiv:2501.16608}
}
read the original abstract

Gait recognition is an emerging identification technology that distinguishes individuals at long distances by analyzing individual walking patterns. Traditional techniques rely heavily on large-scale labeled datasets, which incurs high costs and significant labeling challenges. Recently, researchers have explored unsupervised gait recognition with clustering-based unsupervised domain adaptation methods and achieved notable success. However, these methods directly use pseudo-label generated by clustering and neglect pseudolabel noise caused by domain differences, which affects the effect of the model training process. To mitigate these issues, we proposed a novel model called GaitDCCR, which aims to reduce the influence of noisy pseudo labels on clustering and model training. Our approach can be divided into two main stages: clustering and training stage. In the clustering stage, we propose Dynamic Cluster Parameters (DCP) and Dynamic Weight Centroids (DWC) to improve the efficiency of clustering and obtain reliable cluster centroids. In the training stage, we employ the classical teacher-student structure and propose Confidence-based Pseudo-label Refinement (CPR) and Contrastive Teacher Module (CTM) to encourage noisy samples to converge towards clusters containing their true identities. Extensive experiments on public gait datasets have demonstrated that our simple and effective method significantly enhances the performance of unsupervised gait recognition, laying the foundation for its application in the real-world. We will release the code at https://github.com/YanSun-github/GaitDCCR upon acceptance.

Figures

Figures reproduced from arXiv: 2501.16608 by the authors.

Figure 1
Figure 1. Visualization of clustering results for the previous model and our proposed model. In the figure, samples of di [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The cluster centroid weighting strategy of previous methods and ours. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The framework of our model. “DA” stands for Data Augmentation, and “EMA” represents Exponential Moving Average, which is the [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The generation process of soft pseudo-labels. “DA sample” is obtained by query sample through data augmentation, which simulates the [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: The visualization of data augmentation on NM and CL conditions. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 4
Figure 4. Figure 4: shows the process of optimizing the one-hot label to the soft label. In the student branch, the confidence matrix is obtained by calculating the similarity with all clusters, and then the confidence-based pseudo label y s i is calculated according to Equation 9. In the…
Figure 6
Figure 6. Figure 6: Comparison of previous fixed small and large probability threshold and our dynamic threshold in terms of F1-score and performance. [PITH_FULL_IMAGE:figures/full_fig_p016_6.png]
Figure 7
Figure 7. Figure 7: Comparison of the previous average centroid and our dynamic weighted centroid in terms of mean square error and performance. [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: (a) Different clustering parameter decay methods in DCP. (b) The accuracy of CL clothing during training [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 46 canonical work pages

  1. [1]

    S. Yu, D. Tan, T. Tan, A framework for evaluating the e ffect of view angle, clothing and carrying condition on gait recognition, in: 18th international conference on pattern recognition (ICPR’06), V ol. 4, IEEE, 2006, pp. 441–444. 17

    work page 2006

  2. [2]

    Parashar, A

    A. Parashar, A. Parashar, A. F. Abate, R. S. Shekhawat, I. Rida, Real-time gait biometrics for surveillance applications: A review, Image and Vision Computing 138 (2023) 104784

    work page 2023

  3. [3]

    Azzakhnini, L

    S. Azzakhnini, L. Ballihi, D. Aboutajdine, Combining facial parts for learning gender, ethnicity, and emotional state based on rgb-d informa- tion, ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM) 14 (1s) (2018) 1–14

    work page 2018

  4. [4]

    Lin, Z.-Y

    Y .-S. Lin, Z.-Y . Liu, Y .-A. Chen, Y .-S. Wang, Y .-L. Chang, W. H. Hsu, xcos: An explainable cosine metric for face verification task, ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM) 17 (3s) (2021) 1–16

    work page 2021

  5. [5]

    L. Qi, L. Wang, J. Huo, Y . Shi, Y . Gao, Greyreid: a novel two-stream deep framework with rgb-grey information for person re-identification, ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM) 17 (1) (2021) 1–22

    work page 2021

  6. [6]

    X. Wang, Y . Wang, W. Li, U-net conditional gans for photo-realistic and identity-preserving facial expression synthesis, ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM) 15 (3s) (2019) 1–23

    work page 2019

  7. [7]

    K. Cao, A. K. Jain, Automated latent fingerprint recognition, IEEE transactions on pattern analysis and machine intelligence 41 (4) (2018) 788–800

    work page 2018

  8. [8]

    Sepas-Moghaddam, F

    A. Sepas-Moghaddam, F. M. Pereira, P. L. Correia, Face recognition: a novel multi-level taxonomy based survey, IET Biometrics 9 (2) (2020) 58–67

    work page 2020

Show all 50 references
  1. [9]

    Nguyen, C

    K. Nguyen, C. Fookes, R. Jillela, S. Sridharan, A. Ross, Long range iris recognition: A survey, Pattern Recognition 72 (2017) 123–143

    2017

  2. [10]

    H. Chao, K. Wang, Y . He, J. Zhang, J. Feng, Gaitset: Cross-view gait recognition through utilizing gait as a deep set, IEEE transactions on pattern analysis and machine intelligence 44 (7) (2021) 3467–3478

    2021

  3. [11]

    B. Lin, S. Zhang, M. Wang, L. Li, X. Yu, Gaitgl: Learning discriminative global-local feature representations for gait recognition, arXiv preprint arXiv:2208.01380 (2022)

    2022 arXiv

  4. [12]

    X. Song, Y . Wang, Y . Huang, C. Shan, Gait recognition via view-aware part-wise attention and multi-scale dilated temporal extractor, Image and Vision Computing 156 (2025) 105464

    2025

  5. [13]

    C. Fan, Y . Peng, C. Cao, X. Liu, S. Hou, J. Chi, Y . Huang, Q. Li, Z. He, Gaitpart: Temporal part-based model for gait recognition, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 14225–14233

    2020

  6. [14]

    P. S. Rao, P. Parida, G. Sahu, S. Dash, A multi-view human gait recognition using hybrid whale and gray wolf optimization algorithm with a random forest classifier, Image and Vision Computing 136 (2023) 104721

    2023

  7. [15]

    Y . Sun, X. Feng, L. Ma, L. Hu, M. Nixon, Trigait: Aligning and fusing skeleton and silhouette gait data via a tri-branch network, in: 2023 IEEE International Joint Conference on Biometrics (IJCB), IEEE, 2023, pp. 1–9

    2023

  8. [16]

    J. P. Singh, S. Jain, S. Arora, U. P. Singh, Vision-based gait recognition: A survey, Ieee Access 6 (2018) 70497–70527

    2018

  9. [17]

    M. Wang, X. Guo, B. Lin, T. Yang, Z. Zhu, L. Li, S. Zhang, X. Yu, Dygait: Exploiting dynamic representations for high-performance gait recognition, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 13424–13433

    2023

  10. [18]

    T. Chai, X. Mei, A. Li, Y . Wang, Silhouette-based view-embeddings for gait recognition under multiple views, in: 2021 IEEE international conference on image processing (ICIP), IEEE, 2021, pp. 2319–2323

    2021

  11. [19]

    Ganin, V

    Y . Ganin, V . Lempitsky, Unsupervised domain adaptation by backpropagation, in: International conference on machine learning, PMLR, 2015, pp. 1180–1189

    2015

  12. [20]

    W. Li, X. Zhu, S. Gong, Harmonious attention network for person re-identification, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 2285–2294

    2018

  13. [21]

    Everingham, L

    M. Everingham, L. Van Gool, C. K. Williams, J. Winn, A. Zisserman, The pascal visual object classes (voc) challenge, International journal of computer vision 88 (2010) 303–338

    2010

  14. [22]

    Zheng, X

    J. Zheng, X. Liu, C. Yan, J. Zhang, W. Liu, X. Zhang, T. Mei, Trand: Transferable neighborhood discovery for unsupervised cross-domain gait recognition, in: 2021 IEEE International Symposium on Circuits and Systems (ISCAS), IEEE, 2021, pp. 1–5

    2021

  15. [23]

    Habib, N

    G. Habib, N. Barzilay, O. Shimshi, R. Ben-Ari, N. Darshan, Watch your pose: Unsupervised domain adaption with pose based triplet selection for gait recognition, arXiv preprint arXiv:2307.06751 (2023)

    2023 arXiv

  16. [24]

    X. Ren, S. Hou, C. Cao, X. Liu, Y . Huang, Unsupervised gait recognition with selective fusion, arXiv preprint arXiv:2303.10772 (2023)

    2023 arXiv

  17. [25]

    K. Ma, Y . Fu, D. Zheng, Y . Peng, C. Cao, Y . Huang, Fine-grained unsupervised domain adaptation for gait recognition, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 11313–11322

    2023

  18. [26]

    Zhang, T

    Y . Zhang, T. Xiang, T. M. Hospedales, H. Lu, Deep mutual learning, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 4320–4328

    2018

  19. [27]

    R. Liao, S. Yu, W. An, Y . Huang, A model-based gait recognition method with body pose and human prior knowledge, Pattern Recognition 98 (2020) 107069

    2020

  20. [28]

    Zhang, H

    S. Zhang, H. Tong, J. Xu, R. Maciejewski, Graph convolutional networks: a comprehensive review, Computational Social Networks 6 (1) (2019) 1–23

    2019

  21. [29]

    Zheng, X

    J. Zheng, X. Liu, S. Wang, L. Wang, C. Yan, W. Liu, Parsing is all you need for accurate gait recognition in the wild, in: Proceedings of the 31st ACM International Conference on Multimedia, 2023, pp. 116–124

    2023

  22. [30]

    Z. Wang, S. Hou, M. Zhang, X. Liu, C. Cao, Y . Huang, S. Xu, Landmarkgait: intrinsic human parsing for gait recognition, in: Proceedings of the 31st ACM International Conference on Multimedia, 2023, pp. 2305–2314

    2023

  23. [31]

    Zheng, X

    J. Zheng, X. Liu, W. Liu, L. He, C. Yan, T. Mei, Gait recognition in the wild with dense 3d representations and a benchmark, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 20228–20237

    2022

  24. [32]

    Y . Peng, K. Ma, Y . Zhang, Z. He, Learning rich features for gait recognition by integrating skeletons and silhouettes, Multimedia Tools and Applications 83 (3) (2024) 7273–7294

    2024

  25. [33]

    Huang, X

    T. Huang, X. Ben, C. Gong, W. Xu, Q. Wu, H. Zhou, Gaitdan: Cross-view gait recognition via adversarial domain adaptation, IEEE Transac- tions on Circuits and Systems for Video Technology (2024)

    2024

  26. [34]

    Jaiswal, V

    N. Jaiswal, V . D. Huan, F. Limanta, K. Shinoda, M. Wakasa, Domain-specific adaptation for enhanced gait recognition in practical scenarios, in: 2024 6th International Conference on Image, Video and Signal Processing, 2024, pp. 8–15

    2024

  27. [35]

    R. Li, Q. Jiao, W. Cao, H.-S. Wong, S. Wu, Model adaptation: Unsupervised domain adaptation without source data, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 9641–9650. 18

    2020

  28. [36]

    Saito, D

    K. Saito, D. Kim, S. Sclaro ff, K. Saenko, Universal domain adaptation through self supervision, Advances in neural information processing systems 33 (2020) 16282–16292

    2020

  29. [37]

    Ren, Y .-H

    C.-X. Ren, Y .-H. Liu, X.-W. Zhang, K.-K. Huang, Multi-source unsupervised domain adaptation via pseudo target domain, IEEE Transactions on Image Processing 31 (2022) 2122–2135

    2022

  30. [38]

    Westfechtel, H.-W

    T. Westfechtel, H.-W. Yeh, D. Zhang, T. Harada, Gradual source domain expansion for unsupervised domain adaptation, in: Proceedings of the IEEE/CVF winter conference on applications of computer vision, 2024, pp. 1946–1955

    2024

  31. [39]

    T. Chen, S. Kornblith, M. Norouzi, G. Hinton, A simple framework for contrastive learning of visual representations, in: International conference on machine learning, PMLR, 2020, pp. 1597–1607

    2020

  32. [40]

    K. He, H. Fan, Y . Wu, S. Xie, R. Girshick, Momentum contrast for unsupervised visual representation learning, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 9729–9738

    2020

  33. [41]

    Grill, F

    J.-B. Grill, F. Strub, F. Altché, C. Tallec, P. Richemond, E. Buchatskaya, C. Doersch, B. Avila Pires, Z. Guo, M. Gheshlaghi Azar, et al., Bootstrap your own latent-a new approach to self-supervised learning, Advances in neural information processing systems 33 (2020) 21271– 21284

    2020

  34. [42]

    Caron, I

    M. Caron, I. Misra, J. Mairal, P. Goyal, P. Bojanowski, A. Joulin, Unsupervised learning of visual features by contrasting cluster assignments, Advances in neural information processing systems 33 (2020) 9912–9924

    2020

  35. [43]

    Z. Dai, G. Wang, W. Yuan, S. Zhu, P. Tan, Cluster contrast for unsupervised person re-identification, in: Proceedings of the Asian conference on computer vision, 2022, pp. 1142–1160

    2022

  36. [44]

    Y . Miao, J. Deng, G. Ding, J. Han, Confidence-guided centroids for unsupervised person re-identification, IEEE Transactions on Information Forensics and Security (2024)

    2024

  37. [45]

    C. Fan, S. Hou, J. Wang, Y . Huang, S. Yu, Learning gait representation from massive unlabelled walking videos: A benchmark, IEEE Transactions on Pattern Analysis and Machine Intelligence (2023)

    2023

  38. [46]

    Takemura, Y

    N. Takemura, Y . Makihara, D. Muramatsu, T. Echigo, Y . Yagi, Multi-view large population gait dataset and its performance evaluation for cross-view gait recognition, IPSJ transactions on Computer Vision and Applications 10 (2018) 1–14

    2018

  39. [47]

    Z. Zhu, X. Guo, T. Yang, J. Huang, J. Deng, G. Huang, D. Du, J. Lu, J. Zhou, Gait recognition in the wild: A benchmark, in: Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 14789–14799

    2021

  40. [48]

    C. Fan, J. Liang, C. Shen, S. Hou, Y . Huang, S. Yu, Opengait: Revisiting gait recognition towards better practicality, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 9707–9716

    2023

  41. [49]

    Zhang, L

    Z. Zhang, L. Tran, X. Yin, Y . Atoum, X. Liu, J. Wan, N. Wang, Gait recognition via disentangled representation learning, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 4710–4719

    2019

  42. [50]

    B. Lin, S. Zhang, F. Bao, Gait recognition with multiple-temporal-scale 3d convolutional neural network, in: Proceedings of the 28th ACM international conference on multimedia, 2020, pp. 3054–3062. 19

    2020

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.