Pith. sign in

REVIEW 5 major objections 5 minor 56 references

Distribution-aware Forgetting Compensation for Exemplar-Free Lifelong Person Re-identification

T0 review · 5 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A lifelong person ReID model that never stores old images and never uses knowledge distillation claims to beat every previous method by learning shared text-driven prompts and consolidating per-domain distributions.

desk verdict A coherent prompt+MoE system for exemplar-free LReID, but the SOTA claim is confounded by an unreported CLIP backbone advantage over likely ResNet baselines. read the letter →

arxiv 2504.15041 v3 pith:EIT6FXIX submitted 2025-04-21 cs.CV cs.AI

Shiben Liu , Huijie Fan , Qiang Wang , Baojie Fan , Yandong Tang , Liangqiong Qu This is my paper · ORCID
classification cs.CVcs.AI
keywords lifelongpersonre-identificationcatastrophicforgettingexemplar-freelearningtext-drivenpromptmixtureofexpertsdistribution-awarecompensationknowledgeconsolidation
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 lifelong person re-identification (LReID) can be done well without the two crutches the field has leaned on: storing old exemplars and distilling knowledge from an old model. The proposed DAFC model instead learns a shared pool of prompt components driven by text descriptions, captures each training domain's distribution with dedicated expert networks, and merges those distributions into a common high-dimensional region. On the standard five-domain benchmark, the authors report seen-domain average mAP/R@1 of 65.6/75.9 and 64.7/76.2 under two training orders, and unseen-domain averages of 63.6/60.5 and 64.1/61.4, exceeding prior methods by at least 9.8/6.6 and 6.4/6.2 percentage points. If true, this matters because it removes the privacy and storage costs of rehearsal while avoiding the error accumulation the paper attributes to knowledge distillation.

What carries the argument

The load-bearing object is the domain-shared prompt pool produced by a Text-driven Prompt Aggregation (TPA) module: text features extracted by a frozen text encoder are passed through a Prompt Generator of linear layers to produce M prompt components, from which a query mechanism selects the top-N key-value pairs for each image. This pool feeds the Distribution-based Awareness and Integration (DAI) module, where per-domain expert networks (Domain-Specific Generators) capture each domain's distribution and a Knowledge Adaptive Adjustment layer weights and sums all expert outputs into a shared high-dimensional region. The third mechanism, Knowledge Consolidation (KCM), combines cross-entropy and triplet losses on both the current-domain and shared representations with a cross-domain cosine-consistency term. The central move is that previous-domain distribution information is preserved by frozen experts and replayed through weighted integration, so no exemplar images or distillation losses are needed.

What would settle it

Redo the comparison by reimplementing the strongest baselines (for example DASK and KRKC) on the same frozen CLIP backbone with the same number of prompt components and expert parameters, but without DAFC's distribution integration; if a baseline with only the CLIP backbone and prompt pool reaches or exceeds 65.6/75.9 seen-domain average mAP/R@1 on order-1, the paper's central claim would be overturned.

Watch

Extended reading notes

Core claim

DAFC's central claim is that catastrophic forgetting in LReID can be compensated by treating each domain's data distribution as a learnable object and consolidating all distributions into a shared representation space, while using text-generated prompts to obtain fine-grained, cross-domain-shared instance features. The model freezes a CLIP vision-language backbone and never revisits old samples. Each new domain gets a frozen Domain-Specific Generator made of expert networks; only the current domain's generator is trained. A Knowledge Adaptive Adjustment step uses the text-driven prompt pool to compute soft weights that blend all expert outputs into a shared representation, so old domains are replayed at the distribution level rather than at the pixel level. A Knowledge Consolidation Mechanism then applies instance-level and cross-domain alignment losses. The experiments report that this design outperforms both rehearsal-based and rehearsal-free LReID methods on seen and unseen domains across two training orders.

Load-bearing premise

The paper assumes the comparison to prior methods is apples-to-apples, but it never states the backbones or pretraining of those methods while DAFC uses a frozen CLIP model, so part of the reported gap could come from the architecture rather than the proposed mechanism.

Editorial extensions

If this is right

  • If DAFC holds, lifelong ReID systems can be deployed in privacy-sensitive settings where storing images of past identities is prohibited, because old knowledge lives in frozen expert weights rather than in exemplar galleries.
  • The model's per-domain expert count grows linearly with the number of seen domains, but training cost stays tied to the current domain, suggesting the scheme scales to long streams without revisiting old data.
  • The reported unseen-domain gains imply that text-driven prompts plus distribution consolidation improve generalization to new camera environments, not just retention on seen ones.
  • Because the approach removes knowledge distillation, it sidesteps the paper's identified failure mode where distillation itself accumulates forgetting step by step.

Reading between the lines

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

  • The reported margins may partly reflect the frozen CLIP backbone rather than the forgetting-compensation mechanism; the paper does not state the backbones or pretraining of the compared methods, so an apples-to-apples reimplementation is needed to isolate DAFC's contribution.
  • A natural extension is to vary the attribute-text generator's vocabulary or use language descriptions from other sources; the prompt pool can only express visual attributes that the text side can name, so richer text should directly raise the ceiling on fine-grained discrimination.
  • If the mechanism generalizes, it should transfer to other continual biometric tasks with similar privacy constraints, such as face or gait recognition.
  • The ablation table suggests each of the four components contributes additively; a testable prediction is that DAI alone (without text-driven prompts) should still protect old domains on simple distribution shifts but degrade on fine-grained identity discrimination.
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 / 5 minor

Summary. The paper proposes DAFC, an exemplar-free and distillation-free lifelong person re-identification method built on frozen CLIP visual and text encoders. The method combines three components: Text-driven Prompt Aggregation (TPA), which uses text features to build a domain-shared prompt pool; Distribution-based Awareness and Integration (DAI), which learns per-domain distributions with dedicated expert networks and consolidates them into a shared representation space; and a Knowledge Consolidation Mechanism (KCM) with instance-level discrimination and cross-domain consistency alignment losses. The authors evaluate on five seen domains and six unseen domains under two training orders, reporting seen-domain average mAP/R@1 of 65.6/75.9 and 64.7/76.2, and unseen-domain averages of 63.6/60.5 and 64.1/61.4, and claim state-of-the-art performance over rehearsal-based and rehearsal-free LReID methods. Code is released.

Significance. If the comparison to prior work were properly controlled, the paper would offer a clean and useful recipe for exemplar-free LReID: use frozen CLIP encoders with text-driven prompts to learn shared instance-level representations, and freeze per-domain expert generators to encode domain distributions without storing old exemplars or using knowledge distillation. The paper has clear strengths: the code is released, the component ablations in Table IV isolate the three main modules, and Tables V-VII provide sensitivity analysis for three hyperparameters. The evaluation also covers a standard benchmark with two training orders and six unseen domains. However, the central claim is empirical, and the evidence is currently weakened by a backbone confound in the SOTA comparison, by hyperparameter selection using the same metrics that are subsequently reported, and by internal inconsistencies in the reported gains and loss definitions. The significance of the method is therefore not yet established as stated.

major comments (5)
  1. [Section IV.B, Tables II-III] The SOTA comparison is not controlled for the visual encoder. Section IV.A states that DAFC uses a pre-trained CLIP visual encoder and text encoder that are both frozen, but the paper never reports the backbone or pretraining of the twelve compared methods. Standard LReID baselines such as DKP, DASK, and AKA are typically built on ImageNet-pretrained ResNet-50, which is a substantially weaker feature extractor than a CLIP ViT. The reported gains of roughly 6-10 mAP over DASK and larger gains over the rehearsal-based methods could therefore be explained by the encoder choice rather than by TPA, DAI, or KCM. The internal ablations in Table IV all use the same CLIP backbone and do not isolate this confound. Please add same-backbone comparisons for at least the rehearsal-free baselines, or otherwise disclose and control for the backbone, and temper the claim that DAFC "outperforms state-of-the-art" accordingly.
  2. [Section IV.D, Tables V-VII] Hyperparameters M, N, and k are selected by maximizing the same Seen-Avg and Unseen-Avg metrics that are later reported in Tables II-III. This constitutes selection on the test and unseen-domain evaluation sets, which can inflate the reported numbers and makes the final configuration a result of peeking at the benchmark rather than a fixed design choice. The paper should use a validation split or held-out domains for hyperparameter selection, or alternatively report results for all settings of M/N/k and explicitly state that the reported configuration was chosen without reference to the final evaluation metrics.
  3. [Section III.F, Eqs. (12)-(13)] The domain consistency alignment loss is not well defined as written. Eq. (12) lists M_{s-(s-1)} = gamma(D^s, D^{s-1}), ..., M_{s-1} = gamma(D^s, D^1), but the left-hand indices do not match the right-hand pairs and no index i remains in the expressions. Eq. (13) then computes L_DCA as an average of mu(M_{s-(s-1)}) over i, where mu is called a mean function and gamma is called cosine distance; for s=2 the expression reduces to a single scalar and the summation over i is meaningless. Since L_DCA is an explicit term in the final loss in Eq. (14), the method cannot be reproduced from the current text.
  4. [Abstract and Section IV.B] The claimed improvements are internally inconsistent. The abstract states gains of "9.8%/6.6% and 6.4%/6.2%" over the two training orders, but Tables II-III show that the best rehearsal-free baseline DASK achieves seen-avg mAP/R@1 of 55.4/69.3 on order-1 and 57.1/70.0 on order-2; the corresponding gains over DASK are 10.2/6.6 and 7.6/6.2, not 9.8/6.6 and 6.4/6.2. In addition, the per-domain discussion in Section IV.B lists a negative difference of -2.0 mAP and -4.7 R@1 on MSMT17 V2 order-1 while simultaneously claiming that DAFC "significantly outperforms" the rehearsal-free methods. These statements need to be corrected and quantified against a clearly named baseline.
  5. [Section III.B and Fig. 2] The central text-driven component depends on an "Attribute-Text Generator" taken from the authors' prior work [38], but the manuscript gives no description, training details, or independent validation of this generator, and [38] is an arXiv preprint rather than a peer-reviewed archival reference. Because TPA constructs its prompt pool from the text features produced by this generator, and Table IV attributes a drop of about 6 mAP to removing the Prompt Generator, the contribution is not self-contained and the reader cannot assess whether the reported gains depend on an unvalidated external component. Please specify the generator architecture, how it is obtained, and its failure modes, or include it in the method description.
minor comments (5)
  1. [Section IV.D, Table VI] The text states that "setting N=3 yields the best performance," but Table VI varies N over 8, 10, 12, and 14, and the best row is N=10; the stated value is inconsistent with the table.
  2. [Section IV.C, Table IV] The acronym for the cross-domain consistency component is inconsistent: Section III.F calls it "DCA", Section IV.C defines "Cross-Domain Consistency Alignment (CCA)" and then refers to "w/o CAA", while Table IV labels the row "W/o CCA". Please use a single acronym and ensure the ablation description matches the table.
  3. [Section IV.B, Fig. 4] The anti-forgetting discussion mentions "PAEMA" as a compared method, but PAEMA is not listed among the methods in Tables II-III and is not described in the paper; please clarify what PAEMA is or remove the mention.
  4. [Section IV.A] The implementation details do not specify which CLIP variant is used (e.g., ViT-B/16, ViT-B/32, or ViT-L/14) or how the positional embeddings are handled when input images are resized to 256x128 rather than the CLIP-native resolution. This information is necessary for reproducibility.
  5. [General] There are numerous typographical errors and incomplete sentences, including "repesctively" in Eq. (6), "domian" and "reginon" in Section III.E, the dangling phrase "Ms-(s-1) indicates ." after Eq. (13), and duplicate references [2] and [3] listing the same paper. The manuscript should be carefully proofread.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: DAFC's claims are empirical comparisons against external benchmarks, and no equation reduces to a fitted parameter or to a self-citation chain.

full rationale

DAFC's central claim is an empirical performance comparison (Tables II and III) on a fixed benchmark with held-out unseen domains (VIPeR, GRID, CUHK02, Occ Duke, Occ REID, PRID2011). The proposed losses (Eqs. 8-14) combine standard cross-entropy and triplet terms on the current-domain and aggregated representations, and no evaluation quantity is defined in terms of a fitted parameter or constructed to equal an input. The ablations (Table IV) and hyperparameter sweeps (Tables V-VII) vary architectural choices and report performance; selecting M=20, N=10, and k=5 is ordinary hyperparameter selection, not a fitted 'prediction.' The only self-citation entering the pipeline is the Attribute-Text Generator [38], imported as a text-source module; the paper does not derive DAFC's output from [38]'s output, and the final numbers are measured against external datasets, so this reuse is a reproducibility/credit concern rather than a circular reduction. The backbone mismatch with compared methods is an experimental-fairness concern, not a circularity, because the comparison is externally falsifiable rather than built into the method's equations.

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

The method's central claim rests on empirical assumptions: the attribute-text generator [38] produces useful visual-attribute text, frozen CLIP encoders transfer to person ReID, and frozen per-domain expert networks continue to represent old domains. Hyperparameters M, N, k are fitted on the benchmark and constitute free parameters. No new physical entities are introduced; all proposed components are network modules validated only within the paper.

free parameters (4)
  • M (number of prompt components in domain-shared pool) = 20
    This value was selected via Table V using both seen and unseen metrics.
  • N (top-N key-value pairs) = 10
    This value was selected via Table VI; the text erroneously says N=3.
  • k (expert networks per domain-specific generator) = 5
    This value was selected via Table VII as the best balance; k=6 yields higher seen R@1 but lower unseen metrics.
  • Triplet loss margin m = not reported
    The margin is used in Eq. 9 and Eq. 11, but its value is not specified.
assumptions (4)
  • domain assumption The attribute-text generator [38] produces text features carrying discriminative visual attributes for each instance.
    This is the foundation of TPA; the generator is borrowed from the authors' prior work and is not described or independently validated in this paper.
  • domain assumption Frozen CLIP visual and text encoders provide suitable, generalizable features for person ReID across all domains.
    The entire pipeline depends on CLIP pretraining; no comparison against other backbones is provided.
  • domain assumption Domain-specific generators trained on earlier domains remain frozen and accurately represent old domain distributions during later training.
    DAI freezes previous DSGs (Section III.E) and uses them to compute Dall; if these representations drift or are insufficient, the consolidation fails.
  • ad hoc to paper The softmax weighting over the mean of the domain-shared pool (Eq. 6) yields appropriate integration weights for expert representations.
    This weighting scheme is a design choice introduced without theoretical or empirical justification beyond the final results.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Distribution-aware Forgetting Compensation for Exemplar-Free Lifelong Person Re-identification." pith.science (2026). https://pith.science/paper/EIT6FXIX

@misc{pith2026250415041,
  author       = {Pith},
  title        = {Pith review of: Distribution-aware Forgetting Compensation for Exemplar-Free Lifelong Person Re-identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EIT6FXIX}},
  note         = {Machine review of arXiv:2504.15041}
}
read the original abstract

Lifelong Person Re-identification (LReID) suffers from a key challenge in preserving old knowledge while adapting to new information. The existing solutions include rehearsal-based and rehearsal-free methods to address this challenge. Rehearsal-based approaches rely on knowledge distillation, continuously accumulating forgetting during the distillation process. Rehearsal-free methods insufficiently learn the distribution of each domain, leading to forgetfulness over time. To solve these issues, we propose a novel Distribution-aware Forgetting Compensation (DAFC) model that explores cross-domain shared representation learning and domain-specific distribution integration without using old exemplars or knowledge distillation. We propose a Text-driven Prompt Aggregation (TPA) that utilizes text features to enrich prompt elements and guide the prompt model to learn fine-grained representations for each instance. This can enhance the differentiation of identity information and establish the foundation for domain distribution awareness. Then, Distribution-based Awareness and Integration (DAI) is designed to capture each domain-specific distribution by a dedicated expert network and adaptively consolidate them into a shared region in high-dimensional space. In this manner, DAI can consolidate and enhance cross-domain shared representation learning while alleviating catastrophic forgetting. Furthermore, we develop a Knowledge Consolidation Mechanism (KCM) that comprises instance-level discrimination and cross-domain consistency alignment strategies to facilitate model adaptive learning of new knowledge from the current domain and promote knowledge consolidation learning between acquired domain-specific distributions, respectively. Experimental results show that our DAFC outperforms state-of-the-art methods. Our code is available at https://github.com/LiuShiBen/DAFC.

Figures

Figures reproduced from arXiv: 2504.15041 by the authors.

Figure 1
Figure 1. The motivation of the proposed DAFC method. (a) Rehearsal [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall framework of our DAFC. At the current domain [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Generalization tendency on unseen domains. (a) mAP metric, (b) [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Anti-forgetting curves. (a) mAP metric, (b) R@1 metric. After each [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: The t-SNE visualization results of the learned representations on five domains. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 38 canonical work pages

  1. [38]

    Domain consistency representation learning for lifelong person re-identification,

    S. Liu, Q. Wang, H. Fan, W. Ren, B. Fan, and Y . Tang, “Domain consistency representation learning for lifelong person re-identification,”

    work page

  2. [1]

    Dreamt: Diversity enlarged mu- tual teaching for unsupervised domain adaptive person re-identification,

    Y . Tao, J. Zhang, J. Hong, and Y . Zhu, “Dreamt: Diversity enlarged mu- tual teaching for unsupervised domain adaptive person re-identification,” IEEE Transactions on Multimedia , vol. 25, pp. 4586–4597, 2022

    work page 2022

  3. [2]

    Dual consistency-constrained learning for unsupervised visible-infrared person re-identification,

    B. Yang, J. Chen, C. Chen, and M. Ye, “Dual consistency-constrained learning for unsupervised visible-infrared person re-identification,” IEEE Transactions on Information Forensics and Security , vol. 19, pp. 1767– 1779, 2023

    work page 2023

  4. [3]

    Dual consistency-constrained learning for unsupervised visible- infrared person re-identification,

    ——, “Dual consistency-constrained learning for unsupervised visible- infrared person re-identification,” IEEE Transactions on Information Forensics and Security, vol. 19, pp. 1767–1779, 2023

    work page 2023

  5. [4]

    Anti-forgetting adaptation for unsupervised person re-identification,

    H. Chen, F. Bremond, N. Sebe, and S. Zhang, “Anti-forgetting adaptation for unsupervised person re-identification,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2024

    work page 2024

  6. [5]

    A memorizing and generaliz- ing framework for lifelong person re-identification,

    N. Pu, Z. Zhong, N. Sebe, and M. S. Lew, “A memorizing and generaliz- ing framework for lifelong person re-identification,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, no. 11, pp. 13 567–13 585, 2023

    work page 2023

  7. [6]

    Diverse Representation Embedding for Lifelong Person Re-Identification

    S. Liu, H. Fan, Q. Wang, X. Chen, Z. Han, and Y . Tang, “Diverse representation embedding for lifelong person re-identification,” arXiv preprint arXiv:2403.16003, 2024

    work page Pith review arXiv 2024

  8. [7]

    Cross-modal alternating learning with task-aware representations for continual learning,

    W. Li, B.-B. Gao, B. Xia, J. Wang, J. Liu, Y . Liu, C. Wang, and F. Zheng, “Cross-modal alternating learning with task-aware representations for continual learning,” IEEE Transactions on Multimedia , vol. 26, pp. 5911–5924, 2023

    work page 2023

Show all 56 references
  1. [8]

    Cmoa: Contrastive mixture of adapters for generalized few-shot continual learning,

    Y . Cui, J. Zhao, Z. Yu, R. Cai, X. Wang, L. Jin, A. C. Kot, L. Liu, and X. Li, “Cmoa: Contrastive mixture of adapters for generalized few-shot continual learning,” IEEE Transactions on Multimedia , 2025

    2025

  2. [9]

    Lifelong person re-identification via adaptive knowledge accumulation,

    N. Pu, W. Chen, Y . Liu, E. M. Bakker, and M. S. Lew, “Lifelong person re-identification via adaptive knowledge accumulation,” in CVPR, 2021, pp. 7901–7910

    2021

  3. [10]

    Learning contin- ual compatible representation for re-indexing free lifelong person re- identification,

    Z. Cui, J. Zhou, X. Wang, M. Zhu, and Y . Peng, “Learning contin- ual compatible representation for re-indexing free lifelong person re- identification,” in Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition , 2024, pp. 16 614–16 623

    2024

  4. [11]

    Lstkc: Long short-term knowledge consolidation for lifelong person re-identification,

    K. Xu, X. Zou, and J. Zhou, “Lstkc: Long short-term knowledge consolidation for lifelong person re-identification,” in AAAI, vol. 38, no. 14, 2024, pp. 16 202–16 210

    2024

  5. [12]

    Lifelong person re- identification via knowledge refreshing and consolidation,

    C. Yu, Y . Shi, Z. Liu, S. Gao, and J. Wang, “Lifelong person re- identification via knowledge refreshing and consolidation,” in AAAI, vol. 37, no. 3, 2023, pp. 3295–3303

    2023

  6. [13]

    Exemplar-free lifelong person re-identification via prompt-guided adaptive knowledge consolidation,

    Q. Li, K. Xu, Y . Peng, and J. Zhou, “Exemplar-free lifelong person re-identification via prompt-guided adaptive knowledge consolidation,” International Journal of Computer Vision , vol. 132, no. 11, pp. 4850– 4865, 2024

    2024

  7. [14]

    Distribution-aware knowl- edge prototyping for non-exemplar lifelong person re-identification,

    K. Xu, X. Zou, Y . Peng, and J. Zhou, “Distribution-aware knowl- edge prototyping for non-exemplar lifelong person re-identification,” in CVPR, 2024, pp. 16 604–16 613

    2024

  8. [15]

    Dask: Distribution rehearsing via adaptive style kernel learning for exemplar-free lifelong person re-identification,

    K. Xu, C. Jiang, P. Xiong, Y . Peng, and J. Zhou, “Dask: Distribution rehearsing via adaptive style kernel learning for exemplar-free lifelong person re-identification,” arXiv preprint arXiv:2412.09224 , 2024

    2024 arXiv

  9. [16]

    Auto-selected knowl- edge adapters for lifelong person re-identification,

    X. Qian, R. Wu, G. Cheng, and J. Han, “Auto-selected knowl- edge adapters for lifelong person re-identification,” arXiv preprint arXiv:2405.19005, 2024

    2024 arXiv

  10. [17]

    Learning to prompt for continual learning,

    Z. Wang, Z. Zhang, C.-Y . Lee, H. Zhang, R. Sun, X. Ren, G. Su, V . Perot, J. Dy, and T. Pfister, “Learning to prompt for continual learning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 139–149

    2022

  11. [18]

    S-prompts learning with pre-trained transformers: An occam’s razor for domain incremental learning,

    Y . Wang, Z. Huang, and X. Hong, “S-prompts learning with pre-trained transformers: An occam’s razor for domain incremental learning,” Ad- vances in Neural Information Processing Systems , vol. 35, pp. 5682– 5695, 2022

    2022

  12. [19]

    Dualprompt: Complementary prompting for rehearsal-free continual learning,

    Z. Wang, Z. Zhang, S. Ebrahimi, R. Sun, H. Zhang, C.-Y . Lee, X. Ren, G. Su, V . Perot, J. Dyet al., “Dualprompt: Complementary prompting for rehearsal-free continual learning,” in European conference on computer vision. Springer, 2022, pp. 631–648

    2022

  13. [20]

    Generalising without forgetting for lifelong person re-identification,

    G. Wu and S. Gong, “Generalising without forgetting for lifelong person re-identification,” in Proceedings of the AAAI conference on artificial intelligence, vol. 35, no. 4, 2021, pp. 2889–2897

    2021

  14. [21]

    Lifelong person re-identification by pseudo task knowledge preserva- tion,

    W. Ge, J. Du, A. Wu, Y . Xian, K. Yan, F. Huang, and W.-S. Zheng, “Lifelong person re-identification by pseudo task knowledge preserva- tion,” in AAAI, vol. 36, no. 1, 2022, pp. 688–696

    2022

  15. [22]

    Patch-based knowledge distillation for lifelong person re-identification,

    Z. Sun and Y . Mu, “Patch-based knowledge distillation for lifelong person re-identification,” in ACM MM, 2022, pp. 696–707

    2022

  16. [23]

    Lifelong visible-infrared person re-identification via a tri-token transformer with a query-key mechanism,

    Y . Xing, G. Xiao, M. S. Lew, and S. Wu, “Lifelong visible-infrared person re-identification via a tri-token transformer with a query-key mechanism,” in Proceedings of the 2024 International Conference on Multimedia Retrieval, 2024, pp. 988–997

    2024

  17. [24]

    Learning consistent region features for lifelong person re- identification,

    J. Huang, X. Yu, D. An, Y . Wei, X. Bai, J. Zheng, C. Wang, and J. Zhou, “Learning consistent region features for lifelong person re- identification,” Pattern Recognition, vol. 144, p. 109837, 2023

    2023

  18. [25]

    Scaling vision with sparse mixture of experts,

    C. Riquelme, J. Puigcerver, B. Mustafa, M. Neumann, R. Jenatton, A. Susano Pinto, D. Keysers, and N. Houlsby, “Scaling vision with sparse mixture of experts,” Advances in Neural Information Processing Systems, vol. 34, pp. 8583–8595, 2021

    2021

  19. [26]

    Multimodal contrastive learning with limoe: the language-image mix- ture of experts,

    B. Mustafa, C. Riquelme, J. Puigcerver, R. Jenatton, and N. Houlsby, “Multimodal contrastive learning with limoe: the language-image mix- ture of experts,” Advances in Neural Information Processing Systems , vol. 35, pp. 9564–9576, 2022

    2022

  20. [27]

    Deepseekmoe: Towards ultimate expert spe- cialization in mixture-of-experts language models,

    D. Dai, C. Deng, C. Zhao, R. Xu, H. Gao, D. Chen, J. Li, W. Zeng, X. Yu, Y . Wu et al. , “Deepseekmoe: Towards ultimate expert spe- cialization in mixture-of-experts language models,” arXiv preprint arXiv:2401.06066, 2024

    2024 arXiv

  21. [28]

    Moe-llava: Mixture of experts for large vision-language models,

    B. Lin, Z. Tang, Y . Ye, J. Cui, B. Zhu, P. Jin, J. Huang, J. Zhang, Y . Pang, M. Ning et al., “Moe-llava: Mixture of experts for large vision-language models,” arXiv preprint arXiv:2401.15947 , 2024

    2024 arXiv

  22. [29]

    Llama-moe: Building mixture-of-experts from llama with continual pre- training,

    T. Zhu, X. Qu, D. Dong, J. Ruan, J. Tong, C. He, and Y . Cheng, “Llama-moe: Building mixture-of-experts from llama with continual pre- training,” in Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing , 2024, pp. 15 913–15 923

    2024

  23. [30]

    Boosting continual learning of vision-language models via mixture-of-experts adapters,

    J. Yu, Y . Zhuge, L. Zhang, P. Hu, D. Wang, H. Lu, and Y . He, “Boosting continual learning of vision-language models via mixture-of-experts adapters,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 23 219–23 230

    2024

  24. [31]

    Mixture of experts meets prompt-based continual learning,

    M. Le, H. Nguyen, T. Nguyen, T. Pham, L. Ngo, N. Ho et al., “Mixture of experts meets prompt-based continual learning,” Advances in Neural Information Processing Systems , vol. 37, pp. 119 025–119 062, 2024

    2024

  25. [32]

    Isolation and impartial aggregation: A paradigm of incremental learning without interference,

    Y . Wang, Z. Ma, Z. Huang, Y . Wang, Z. Su, and X. Hong, “Isolation and impartial aggregation: A paradigm of incremental learning without interference,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 8, 2023, pp. 10 209–10 217

    2023

  26. [33]

    Progressive prompts: Continual learning for language models,

    A. Razdaibiedina, Y . Mao, R. Hou, M. Khabsa, M. Lewis, and A. Alma- hairi, “Progressive prompts: Continual learning for language models,” arXiv preprint arXiv:2301.12314 , 2023

    2023 arXiv

  27. [34]

    Introducing language guidance in prompt-based continual learning,

    M. G. Z. A. Khan, M. F. Naeem, L. Van Gool, D. Stricker, F. Tombari, and M. Z. Afzal, “Introducing language guidance in prompt-based continual learning,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 11 463–11 473

    2023

  28. [35]

    Consistent prompting for rehearsal- free continual learning,

    Z. Gao, J. Cen, and X. Chang, “Consistent prompting for rehearsal- free continual learning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 28 463–28 473

    2024

  29. [36]

    Coda-prompt: Contin- ual decomposed attention-based prompting for rehearsal-free continual learning,

    J. S. Smith, L. Karlinsky, V . Gutta, P. Cascante-Bonilla, D. Kim, A. Arbelle, R. Panda, R. Feris, and Z. Kira, “Coda-prompt: Contin- ual decomposed attention-based prompting for rehearsal-free continual learning,” in CVPR, 2023, pp. 11 909–11 919

    2023

  30. [37]

    Convolutional prompting meets language models for continual learning,

    A. Roy, R. Moulick, V . K. Verma, S. Ghosh, and A. Das, “Convolutional prompting meets language models for continual learning,” in Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 23 616–23 626

    2024

  31. [39]

    Scalable person re-identification: A benchmark,

    L. Zheng, L. Shen, L. Tian, S. Wang, J. Wang, and Q. Tian, “Scalable person re-identification: A benchmark,” in Proceedings of the IEEE international conference on computer vision , 2015, pp. 1116–1124

    2015

  32. [40]

    End-to-end deep learning for person search,

    T. Xiao, S. Li, B. Wang, L. Lin, and X. Wang, “End-to-end deep learning for person search,” arXiv preprint arXiv:1604.01850, vol. 2, no. 2, p. 4, 2016. 11

    2016 arXiv

  33. [41]

    Performance measures and a data set for multi-target, multi-camera tracking,

    E. Ristani, F. Solera, R. Zou, R. Cucchiara, and C. Tomasi, “Performance measures and a data set for multi-target, multi-camera tracking,” in European conference on computer vision . Springer, 2016, pp. 17–35

    2016

  34. [42]

    Person transfer gan to bridge domain gap for person re-identification,

    L. Wei, S. Zhang, W. Gao, and Q. Tian, “Person transfer gan to bridge domain gap for person re-identification,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 79– 88

    2018

  35. [43]

    Deepreid: Deep filter pairing neural network for person re-identification,

    W. Li, R. Zhao, T. Xiao, and X. Wang, “Deepreid: Deep filter pairing neural network for person re-identification,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2014, pp. 152– 159

    2014

  36. [44]

    Viewpoint invariant pedestrian recognition with an ensemble of localized features,

    D. Gray and H. Tao, “Viewpoint invariant pedestrian recognition with an ensemble of localized features,” in Computer Vision–ECCV 2008: 10th European Conference on Computer Vision, Marseille, France, October 12-18, 2008, Proceedings, Part I 10 . Springer, 2008, pp. 262–275

    2008

  37. [45]

    Time-delayed correlation analysis for multi-camera activity understanding,

    C. C. Loy, T. Xiang, and S. Gong, “Time-delayed correlation analysis for multi-camera activity understanding,” International Journal of Computer Vision, vol. 90, pp. 106–129, 2010

    2010

  38. [46]

    Locally aligned feature transforms across views,

    W. Li and X. Wang, “Locally aligned feature transforms across views,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2013, pp. 3594–3601

    2013

  39. [47]

    Pose-guided feature alignment for occluded person re-identification,

    J. Miao, Y . Wu, P. Liu, Y . Ding, and Y . Yang, “Pose-guided feature alignment for occluded person re-identification,” in Proceedings of the IEEE/CVF international conference on computer vision , 2019, pp. 542– 551

    2019

  40. [48]

    Occluded person re- identification,

    J. Zhuo, Z. Chen, J. Lai, and G. Wang, “Occluded person re- identification,” in 2018 IEEE International Conference on Multimedia and Expo (ICME) . IEEE, 2018, pp. 1–6

    2018

  41. [49]

    Person re- identification by descriptive and discriminative classification,

    M. Hirzer, C. Beleznai, P. M. Roth, and H. Bischof, “Person re- identification by descriptive and discriminative classification,” in Image Analysis: 17th Scandinavian Conference, SCIA 2011, Ystad, Sweden, May 2011. Proceedings 17 . Springer, 2011, pp. 91–102

    2011

  42. [50]

    Learning transferable visual models from natural language supervision,

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

    2021

  43. [51]

    A method for stochastic optimization,

    D. Kinga, J. B. Adam et al. , “A method for stochastic optimization,” in International conference on learning representations (ICLR) , vol. 5, no. 6. San Diego, California;, 2015

    2015

  44. [52]

    Similarity-preserving knowledge distillation,

    F. Tung and G. Mori, “Similarity-preserving knowledge distillation,” in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 1365–1374

    2019

  45. [53]

    Learning without forgetting,

    Z. Li and D. Hoiem, “Learning without forgetting,” IEEE transactions on pattern analysis and machine intelligence , vol. 40, no. 12, pp. 2935– 2947, 2017

    2017

  46. [54]

    Continual repre- sentation learning for biometric identification,

    B. Zhao, S. Tang, D. Chen, H. Bilen, and R. Zhao, “Continual repre- sentation learning for biometric identification,” in Proceedings of the IEEE/CVF winter conference on applications of computer vision , 2021, pp. 1198–1208. Shiben Liu received his B.E. and M.S. degrees in Ele...

    2021

  47. [2014]

    Her research interests include deep learning on image processing and medical image processing and applications

    She is currently a Research Scientist with the Institute of Shenyang Automation of the Chinese Academy of Sciences. Her research interests include deep learning on image processing and medical image processing and applications. Qiang Wang received the B.E. and M.S. degrees in ...

    2004

  48. [2024]

    Available: https://arxiv.org/abs/2409.19954

    [Online]. Available: https://arxiv.org/abs/2409.19954

    arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.