DASK: Distribution Rehearsing via Adaptive Style Kernel Learning for Exemplar-Free Lifelong Person Re-Identification

Chenghao Jiang; Jiahuan Zhou; Kunlun Xu; Peixi Xiong; Yuxin Peng

REVIEW 3 major objections 4 minor 59 references

DASK: Distribution Rehearsing via Adaptive Style Kernel Learning for Exemplar-Free Lifelong Person Re-Identification

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read DASK claims that an exemplar-free lifelong person re-identification method can rehearse past domain distributions by synthesizing old-style images from new data, beating existing exemplar-free methods by 3.6%–6.8% in anti-forgetting and…

desk verdict A genuinely new rehearsing mechanism with solid empirical gains, but the load-bearing distribution-transfer claim is unverified and one ablation quietly undermines it. read the letter →

arxiv 2412.09224 v3 pith:UXFM4KOP submitted 2024-12-12 cs.CV

Kunlun Xu , Chenghao Jiang , Peixi Xiong , Yuxin Peng , Jiahuan Zhou This is my paper · ORCID

classification cs.CV

keywords lifelongpersonre-identificationexemplar-freecatastrophicforgettingdistributionrehearsaladaptivekernelpredictionstyletransferknowledgeconsolidationdomaingeneralization

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 tries to establish that lifelong person re-identification can be made exemplar-free without giving up the anti-forgetting power of data replay: instead of storing old images, a model can learn to restyle new images into the old domain's look and train on those synthetic old-style images. The proposed method, DASK, predicts a per-image convolution kernel (a distribution transfer kernel) that converts new-domain images into convincing old-domain style, and then jointly trains on real new data and generated old-style data under a knowledge-consolidation loss. The authors report consistent gains over existing exemplar-free methods: 3.6%–6.8% better average mAP/Rank-1 on seen domains and 4.5%–6.5% on unseen domains across two training orders. If correct, this gives a privacy-preserving route to lifelong ReID that approaches replay-based methods on seen domains and exceeds them on generalization.

What carries the argument

The load-bearing object is the Distribution Transfer Kernel, a C×C×k×k convolution kernel predicted per image by AKPNet and applied as $k_i \circledast x_i$ to restyle a new image into a target domain's style. Because each image gets its own kernel, the model can handle instance-specific color and texture offsets, which the authors show is a strict generalization of statistical color-transfer methods that only predict per-channel mean and standard deviation. This kernel, trained via a self-supervised reconstruction loss on synthetically augmented data, is what makes old-domain rehearsal possible without exemplars.

What would settle it

Compute a distribution-distance metric (e.g., FID or a domain classifier) between real old-domain images and DASK's generated old-style images at each training step; if the distance is comparable to the gap between two real domains, the rehearsal signal is likely too weak to explain the anti-forgetting gains, which would instead stem from the extra data augmentation.

Watch

Extended reading notes

Core claim

DASK's central claim is that you can 'rehearse' a past data distribution without any stored exemplars by learning, at each step, a generator that inverts synthetic augmentations of the current domain: an Adaptive Kernel Prediction Network (AKPNet) sees an image and outputs an instance-specific convolution kernel; convolving the image with that kernel reconstructs the original from its augmented version, so that later, when a new domain's data arrives, the same old AKPNet can be applied to produce plausible old-style images. These generated images are assigned the identity labels of the new images and fed, together with real new data, into a joint training loss that combines ReID losses and cross-instance similarity distillation. The result, the authors argue, is a better balance between learning the new domain and not forgetting older ones, because the model is continually re-exposed to the old distribution at the input level rather than only through distilled output constraints.

Load-bearing premise

The whole rehearsal relies on the AKPNet trained against synthetic color and blur shifts of one domain being able to transform genuinely new, unseen domain images into convincing old-domain style; the paper shows visual examples but never measures whether the generated images actually match the old distribution.

Editorial extensions

If this is right

If correct, exemplar-free LReID can match or exceed replay-based methods on seen-domain retention while avoiding privacy and storage issues.
The same distribution-rehearsal recipe could apply to other lifelong tasks with strong domain gaps, since it only needs a style-transfer learner and a joint consolidation loss.
The method's generalization gains suggest that synthetic old-style data acts as a domain augmentation, teaching the model domain-invariant features.
The single retained AKPNet (instead of all historical models) keeps storage overhead constant, making the approach practical for long task streams.

Reading between the lines

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

A direct quantitative test of generation fidelity (e.g., FID or domain classifier accuracy between generated and real old-domain images) is missing from the paper; if such a test showed poor fidelity on large gaps, the anti-forgetting benefits would likely shrink.
The assumption that synthetic augmentations (color statistics plus Gaussian blur) span the space of real domain gaps is untested; future domains with structural or resolution shifts may not be invertible by a single convolution kernel.

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.

Desk Editor's Note

read the letter

Quick take: DASK is a real step forward for exemplar-free LReID. The idea—train a lightweight network to predict a per-instance convolution kernel that restyles new data into an old domain's look, then use those synthetic old-style images for rehearsal—is new relative to the KD and replay literature, and the authors execute it carefully. The gains over the best exemplar-free methods (3.6–6.8% on seen domains, 4.5–6.5% on unseen) are consistent across two training orders, and the paper includes a reasonable ablation suite and releases code. The comparisons to replay-based methods, both with and without replay, are a useful addition.

The soft spots are real but not disqualifying. First, the central assumption—that an AKPNet trained to invert synthetic color-and-blur augmentations will map arbitrary future-domain images to a convincing past-domain style—is never tested quantitatively. The paper shows t-SNE and sample images, but no FID, no domain-classifier accuracy, no human study. Without one of those, we cannot tell whether the generated data actually represents the old distribution or is just a mild style jitter. Second, the appendix ablation on preserving multiple AKPNet models gave me pause: using all historical AKPNet models to rehearse every past domain yielded no improvement over retaining only the most recent one. If the mechanism were faithfully rehearsing each old domain, older domains should benefit from their own generators. The paper's explanation—that rehearsal strengthens features from previous steps—does not address that. This suggests the real effect may be generic multi-style augmentation rather than per-domain rehearsal. Third, all numbers are single-run point estimates; no error bars anywhere.

None of this destroys the empirical result. The method works better than the alternatives, and the mechanism is plausible even if the story is oversold. But the authors should be asked to verify the transfer fidelity and to reconcile the multi-AKPNet ablation with their framing. With those additions this would be a strong paper; without them, the contribution is still solid but less certain than the abstract implies.

For the record: I'd send this to reviewers. It deserves a serious referee, and the requests are concrete and addressable.

Referee Report

3 major / 4 minor

Summary. The paper proposes DASK, an exemplar-free lifelong person re-identification method. At each training step, a Distribution Rehearser Learning (DRL) mechanism trains an Adaptive Kernel Prediction Network (AKPNet) to transform distribution-augmented images back into the current domain style, using an instance-specific convolution kernel and a self-supervised reconstruction loss. At the next step, the old AKPNet is applied to the new domain data to generate old-style images, and a joint knowledge consolidation module trains the ReID model with a combination of a classical ReID loss, similarity-preserving distillation, and the same losses on the generated old-style data. Experiments on the LReID benchmark with five seen and seven unseen domains under two training orders report state-of-the-art results, with claimed improvements of 3.6%-6.8% on seen-domain anti-forgetting and 4.5%-6.5% on unseen-domain generalization over existing exemplar-free methods.

Significance. If the central mechanism is valid, DASK is a meaningful contribution: it offers a privacy-preserving alternative to exemplar replay and demonstrates that image-level distribution rehearsal can improve both anti-forgetting and generalization in lifelong ReID. The paper has clear strengths: it releases code, it compares against both replay-based and no-replay baselines, and its component ablations (Tables 3 and 4) support the view that the proposed losses and kernel prediction contribute to the reported gains. The central derivation is not circular: the AKPNet is trained self-supervised on the old domain, not on the final evaluation targets. The main weakness is that the load-bearing assumption of cross-domain transfer fidelity is unvalidated, and one of the paper's own ablations indirectly undercuts the domain-specific rehearsal interpretation. The result therefore needs strengthening before the paper can be accepted.

major comments (3)

[Distribution Rehearser Learning, Eqs. (7)-(8)] The central claim that 'when the new data D_{t+1} from an arbitrary domain is given, the distribution of D_t can be rehearsed' is not quantitatively verified. The AKPNet is trained only to invert color mean/std shifts and Gaussian blur applied to D_{t-1}, but no metric such as FID, a domain-classifier accuracy, or a feature-distance measure compares the generated old-style images from D_t against real D_{t-1} images. The t-SNE overlap in Fig. 5(a) is computed on features of the ReID model trained with those generated images and cannot by itself establish style fidelity. Please add a direct fidelity measurement and, ideally, compare the downstream anti-forgetting gain against the real-exemplar upper bound.
[Appendix, 'Preserving Multiple AKPNet Models', Fig. 11] The ablation reporting that using all historical AKPNet models gives no improvement over retaining only Ψ_{t-1} is in tension with the domain-specific rehearsal hypothesis. If generated data actually rehearsed each past domain, then at step t the older models Ψ_{t-2}, ..., Ψ_1 should allow targeted rehearsal of earlier domains and improve their per-domain retention. The absence of any benefit suggests the generated data may act as generic style augmentation rather than faithful domain-specific replay. Please report per-domain results for this ablation and explain why preserving older AKPNet models provides no gain for older-domain knowledge.
[Tables 1-2 and Tables 5-6] No error bars, repeated runs, or significance tests are reported. The headline improvements are 3.6%-6.8% on seen and 4.5%-6.5% on unseen domains, but individual domain results sometimes go against the average (e.g., Table 1 CUHK-SYSU mAP is 81.9 for DASK versus 83.6 for DKP), making it unclear whether the average gains are stable. Please report mean±std over at least three seeds, or provide per-run values, for the main comparisons.

minor comments (4)

[Eq. (10)] In the reconstruction formula, the sum over j ∈ {r,g,b} multiplies k^j_{i,p,q} by (x^r_i')_{m+p,n+q}, which should presumably be (x^j_i')_{m+p,n+q}; as written, the cross-channel mixing is not represented. In addition, the statement that Eq. (9) is a special case 'when k^j_{i,p,q}=0, w.r.t. p≠0, q≠0' should read 'when all taps with (p,q)≠(0,0) vanish and only the r-channel central tap remains'; the current wording has the condition inverted.
[Experiments section] The first comparison tables are numbered Table 1 and Table 2, but the text refers to them several times as 'Tab. 5 and Tab. 6'; please correct the cross-references.
[Appendix, 'Comparison with Deep Generation Network'] The text contains the typo 'APKNet' (in the sentence 'the APKNet of DASK can generate diverse-style images'); this should be 'AKPNet' throughout to match the defined abbreviation.
[Appendix, Fig. 11] The statement that preserving all historical AKPNet models achieves 'comparable performance' is not quantified in the text; please include the numerical mAP/R@1 values for this ablation rather than only a figure.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the distribution-rehearsal signal comes from a self-supervised AKPNet trained on current-domain data and is evaluated against external LReID methods; the unvalidated arbitrary-domain transfer assumption is a correctness risk, not a circular reduction.

full rationale

The paper's central derivation chain is not circular by its own equations. The AKPNet is trained in DRL with the self-supervised reconstruction loss L_ReC = ||x_i - x''_i|| on the current domain D_t, where x''_i = Psi_t(x'_i) circl x'_i and x'_i is a color-statistic and Gaussian-blur augmentation of x_i (Eqs. 7-8). At step t, DRRT applies the old Psi_{t-1} to new data D_t via Eq. 1 and trains M_t with L*_ReID and L*_SKD. None of these objectives involves the final evaluation metrics, namely seen-domain mAP/R@1 on historical datasets or unseen-domain accuracy; those benchmark targets enter only as external test sets and as comparison methods implemented independently. The self-citations to LSTKC, DKP, and CKP provide a similarity-KD loss and an EMA schedule, but these are standard components and are not used as uniqueness theorems or as evidence that the distribution-rehearsing claim itself holds; they do not make the reported benchmark gains reduce to fitted values. The main weakness is empirical rather than circular: the sentence in Section Distribution Rehearser Learning, 'when the new data D_{t+1} from an arbitrary domain is given, the distribution of D_t can be rehearsed', extrapolates from synthetic augmentations to arbitrary real domain gaps, and the generated old-style data are only shown qualitatively (Figs. 5 and 10), with no FID, domain-classifier accuracy, or other fidelity metric. The appendix ablation 'Preserving Multiple AKPNet Models' also shows no improvement from retaining all historical AKPNet models, which indirectly challenges the faithful per-domain rehearsal interpretation. These are validity and generalization concerns, not circularity: the reported gains are not by construction equal to any fitted input, and the comparisons are against external methods. Accordingly, the appropriate circularity finding is 0.

Assumptions & free parameters 4 free parameters · 3 assumptions · 2 invented entities

The method depends on two design choices that carry no external validation: the synthetic augmentation covering all future domain shifts, and the assumption that restyled data with new labels rehearses old-domain knowledge. The tuned hyperparameters alpha, beta, lambda, and kernel count are fitted to the benchmark. The AKPNet and transfer kernels are internal learned components rather than independently evidenced entities.

free parameters (4)

alpha (relation distillation loss weight) = 1.0
Eq. (5); tuned on the LReID benchmark, ablation in Fig. 6a.
beta (rehearsing-guided loss weight) = 4.5
Eq. (5); tuned on the LReID benchmark, ablation in Fig. 6b.
lambda (EMA fusion weight) = 0.5
Eq. (6); tuned on the LReID benchmark, ablation in Fig. 6c.
Nk (number of predicted style kernels) = 1
Appendix; single kernel chosen as optimal in Fig. 8.

assumptions (3)

domain assumption Synthetic color and blur augmentation of the current domain spans the distribution shift to any future domain
Underpins DRL; if false, the old AKPNet cannot restyle new-domain images into old-domain style. Stated in 'when the new data D_{t+1} from an arbitrary domain is given, the distribution of D_t can be rehearsed'.
domain assumption Cross-instance similarity distillation (L_SKD) preserves historical knowledge
Borrowed from prior work; at step t it constrains new model outputs to old model on new and generated data.
domain assumption Training on old-style data with new identity labels consolidates historical discriminative knowledge
Eq. (5) uses L*_ReID on generated old-style data labeled with new identities; the paper's claim that this enhances historical knowledge consolidation is an assumption, since no old identities are replayed.

invented entities (2)

Distribution Transfer Kernel
purpose: Per-instance convolution kernel predicted by AKPNet to transform images between domains (Eq. 1).
A learned internal representation with no external falsifiable prediction tied to it.
AKPNet
purpose: Lightweight network predicting instance-specific transfer kernels for distribution rehearsal.
Proposed model component evaluated only through downstream ReID accuracy.

how reviews work

0 comments

Cite this review

Pith. "Pith review of DASK: Distribution Rehearsing via Adaptive Style Kernel Learning for Exemplar-Free Lifelong Person Re-Identification." pith.science (2026). https://pith.science/paper/UXFM4KOP

@misc{pith2026241209224,
  author       = {Pith},
  title        = {Pith review of: DASK: Distribution Rehearsing via Adaptive Style Kernel Learning for Exemplar-Free Lifelong Person Re-Identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UXFM4KOP}},
  note         = {Machine review of arXiv:2412.09224}
}

read the original abstract

Lifelong person re-identification (LReID) is an important but challenging task that suffers from catastrophic forgetting due to significant domain gaps between training steps. Existing LReID approaches typically rely on data replay and knowledge distillation to mitigate this issue. However, data replay methods compromise data privacy by storing historical exemplars, while knowledge distillation methods suffer from limited performance due to the cumulative forgetting of undistilled knowledge. To overcome these challenges, we propose a novel paradigm that models and rehearses the distribution of the old domains to enhance knowledge consolidation during the new data learning, possessing a strong anti-forgetting capacity without storing any exemplars. Specifically, we introduce an exemplar-free LReID method called Distribution Rehearsing via Adaptive Style Kernel Learning (DASK). DASK includes a Distribution Rehearser Learning (DRL) mechanism that learns to transform arbitrary distribution data into the current data style at each learning step. To enhance the style transfer capacity of DRL, an Adaptive Kernel Prediction Network (AKPNet) is explored to achieve an instance-specific distribution adjustment. Additionally, we design a Distribution Rehearsing-driven LReID Training (DRRT) module, which rehearses old distribution based on the new data via the old AKPNet model, achieving effective new-old knowledge accumulation under a joint knowledge consolidation scheme. Experimental results show our DASK outperforms the existing methods by 3.6%-6.8% and 4.5%-6.5% on anti-forgetting and generalization capacity, respectively. Our code is available at https://github.com/zhoujiahuan1991/AAAI2025-LReID-DASK

Figures

Figures reproduced from arXiv: 2412.09224 by the authors.

**Figure 2.** The overall components of our DASK method. At each training step, a new domain dataset [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Anti-forgetting tendency on seen domains. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 5.** Figure 5: Visualization of distribution rehearsing effects. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

**Figure 6.** Figure 6: Ablation studies on the hyperparameters in this paper. The values marked by the dashed lines are adopted by our [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

**Figure 7.** Figure 7: Variants of predictable style transfer kernel designs. (a) Predicting a single-layer style transfer kernel. (b) Predicting [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

**Figure 8.** Figure 8: Results comparison of Baseline and DASK with [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]

**Figure 11.** Figure 11: Ablation on different numbers of preserved his [PITH_FULL_IMAGE:figures/full_fig_p011_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 54 canonical work pages

[1]

, " * write output.state after.block = add.period write newline

ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...
[2]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...
[3]

Branch, H. O. S. D. 2006. Imagery Library for Intelligent Detection Systems (i-lids). In 2006 IET Conf. on Crime and Security, 445--448. IET

work page 2006
[4]

Chen, H.; Lagadec, B.; and Bremond, F. 2022. Unsupervised Lifelong Person Re-identification via Contrastive Rehearsal. arXiv:2203.06468

arXiv 2022
[5]

Cui, Z.; Zhou, J.; Wang, X.; Zhu, M.; and Peng, Y. 2024. Learning Continual Compatible Representation for Re-indexing Free Lifelong Person Re-identification. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 16614--16623. IEEE

work page 2024
[6]

Deng, W.; Zheng, L.; Ye, Q.; Kang, G.; Yang, Y.; and Jiao, J. 2018. Image-Image Domain Adaptation with Preserved Self-Similarity and Domain-Dissimilarity for Person Re-Identification. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 994--1003. IEEE

work page 2018
[7]

Ge, W.; Du, J.; Wu, A.; Xian, Y.; Yan, K.; Huang, F.; and Zheng, W.-S. 2022. Lifelong Person Re-identification by Pseudo Task Knowledge Preservation. In Proc. AAAI Conf. Artif. Intell., 688--696

work page 2022
[8]

Gray, D.; and Tao, H. 2008. Viewpoint Invariant Pedestrian Recognition with an Ensemble of Localized Features. In Proc. Eur. Conf. Comput. Vis., 262--275. Springer

work page 2008

Show all 59 references

[9]

Gu, J.; Luo, H.; Wang, K.; Jiang, W.; You, Y.; and Zhao, J. 2023. Color Prompting for Data-free Continual Unsupervised Domain Adaptive Person Re-Identification. arXiv preprint arXiv:2308.10716

2023 arXiv
[10]

He, S.; Luo, H.; Wang, P.; Wang, F.; Li, H.; and Jiang, W. 2021. TransReID: Transformer-based Object Re-Identification. In Proc. IEEE Int. Conf. Comput. Vis., 14993--15002. IEEE

2021
[11]

M.; and Bischof, H

Hirzer, M.; Beleznai, C.; Roth, P. M.; and Bischof, H. 2011. Person Re-Identification by Descriptive and Discriminative Classification. In Image Analysis, 91--102. Springer

2011
[12]

Hong, K.; Jeon, S.; Yang, H.; Fu, J.; and Byun, H. 2021. Domain-aware universal style transfer. In Proc. IEEE Int. Conf. Comput. Vis., 14609--14617. IEEE

2021
[13]

Hong, M.; Choi, J.; and Kim, G. 2021. Stylemix: Separating Content and Style for Enhanced Data Augmentation. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 14862--14870. IEEE

2021
[14]

Howard, A.; Sandler, M.; Chu, G.; Chen, L.-C.; Chen, B.; Tan, M.; Wang, W.; Zhu, Y.; Pang, R.; Vasudevan, V.; et al. 2019. Searching for Mobilenetv3. In Proc. IEEE Int. Conf. Comput. Vis., 1314--1324. IEEE

2019
[15]

Huang, J.; Yu, X.; An, D.; Wei, Y.; Bai, X.; Zheng, J.; Wang, C.; and Zhou, J. 2023. Learning Consistent Region Features for Lifelong Person Re-Identification. Pattern Recognition, 144: 109837

2023
[16]

Huang, Z.; Zhang, Z.; Lan, C.; Zeng, W.; Chu, P.; You, Q.; Wang, J.; Liu, Z.; and Zha, Z.-j. 2022. Lifelong Unsupervised Domain Adaptive Person Re-identification with Coordinated Anti-forgetting and Adaptation. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 14288--14297. IEEE

2022
[17]

Li, Q.; Peng, Y.; and Zhou, J. 2024. FCS: Feature Calibration and Separation for Non-Exemplar Class Incremental Learning. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 28495--28504. IEEE

2024
[18]

Li, Q.; Xu, K.; Peng, Y.; and Zhou, J. 2024. Exemplar-Free Lifelong Person Re-identification via Prompt-Guided Adaptive Knowledge Consolidation. Int. J. Comput. Vis., 1--16

2024
[19]

Li, W.; and Wang, X. 2013. Locally Aligned Feature Transforms across Views. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 3594--3601. IEEE

2013
[20]

Li, W.; Zhao, R.; and Wang, X. 2012. Human Reidentification with Transferred Metric Learning. In ACCV, 31--44. Springer

2012
[21]

Li, W.; Zhao, R.; Xiao, T.; and Wang, X. 2014. DeepReID: Deep Filter Pairing Neural Network for Person Re-Identification. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 152--159. IEEE

2014
[22]

Li, Y.; Song, J.; Ni, H.; and Shen, H. T. 2023. Style-Controllable Generalized Person Re-Identification. In Proc. ACM Int. Conf. Multimedia, 7912--7921

2023
[23]

Li, Z.; and Hoiem, D. 2017. Learning without Forgetting. IEEE Trans. Pattern Anal. Mach. Intell., 40(12): 2935--2947

2017
[24]

Liu, Z.; Peng, Y.; and Zhou, J. 2024. Compositional Prompting for Anti-Forgetting in Domain Incremental Learning. Int. J. Comput. Vis., 1--18

2024
[25]

C.; Xiang, T.; and Gong, S

Loy, C. C.; Xiang, T.; and Gong, S. 2010. Time-Delayed Correlation Analysis for Multi-Camera Activity Understanding. Int. J. Comput. Vis., 90(1): 106--129

2010
[26]

D.; Mirza, S.; Zakeri, A.; Gupta, A.; Khaldi, K.; Aloui, R.; Mantini, P.; Shah, S

Nguyen, V. D.; Mirza, S.; Zakeri, A.; Gupta, A.; Khaldi, K.; Aloui, R.; Mantini, P.; Shah, S. K.; and Merchant, F. 2024. Tackling Domain Shifts in Person Re-Identification: A Survey and Analysis. In CVPR, 4149--4159. IEEE

2024
[27]

M.; and Lew, M

Pu, N.; Chen, W.; Liu, Y.; Bakker, E. M.; and Lew, M. S. 2021. Lifelong Person Re-Identification via Adaptive Knowledge Accumulation. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 7897--7906. IEEE

2021
[28]

M.; and Lew, M

Pu, N.; Liu, Y.; Chen, W.; Bakker, E. M.; and Lew, M. S. 2022. Meta Reconciliation Normalization for Lifelong Person Re-Identification. In Proc. ACM Int. Conf. Multimedia, 541--549

2022
[29]

Pu, N.; Zhong, Z.; Sebe, N.; and Lew, M. S. 2023. A Memorizing and Generalizing Framework for Lifelong Person Re-Identification. IEEE Trans. Pattern Anal. Mach. Intell

2023
[30]

Rebuffi, S.-A.; Kolesnikov, A.; Sperl, G.; and Lampert, C. H. 2017. iCaRL: Incremental Classifier and Representation Learning. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 5533--5542. IEEE

2017
[31]

Ristani, E.; Solera, F.; Zou, R.; Cucchiara, R.; and Tomasi, C. 2016. Performance Measures and A Data Set for Multi-Target, Multi-Camera tracking. In Proc. Eur. Conf. Comput. Vis., 17--35. Springer

2016
[32]

Shi, J.; Yin, X.; Chen, Y.; Zhang, Y.; Zhang, Z.; Xie, Y.; and Qu, Y. 2025. Multi-memory matching for unsupervised visible-infrared person re-identification. In Proc. Eur. Conf. Comput. Vis., 456--474. Springer

2025
[33]

Shi, J.; Yin, X.; Zhang, Y.; Xie, Y.; Qu, Y.; et al. 2024. Learning commonality, divergence and variety for unsupervised visible-infrared person re-identification. In Proc. Int. Conf. Neural Inf. Process. Syst

2024
[34]

Shi, J.; Zhang, Y.; Yin, X.; Xie, Y.; Zhang, Z.; Fan, J.; Shi, Z.; and Qu, Y. 2023. Dual Pseudo-Labels Interactive Self-Training for Semi-Supervised Visible-Infrared Person Re-Identification. In Proc. IEEE Int. Conf. Comput. Vis., 11218--11228

2023
[35]

Shi, W.; and Ye, M. 2023. Prototype Reminiscence and Augmented Asymmetric Knowledge Aggregation for Non-Exemplar Class-Incremental Learning. In Proc. IEEE Int. Conf. Comput. Vis., 1772--1781. IEEE

2023
[36]

Sun, Z.; and Mu, Y. 2022. Patch-Based Knowledge Distillation for Lifelong Person Re-Identification. In Proc. ACM Int. Conf. Multimedia, 696--707

2022
[37]

Tan, W.; Ding, C.; Wang, P.; Gong, M.; and Jia, K. 2023. Style Interleaved Learning for Generalizable Person Re-Identification. IEEE Trans. Multimedia, 26: 1600--1612

2023
[38]

Tung, F.; and Mori, G. 2019. Similarity-Preserving Knowledge Distillation. In Proc. IEEE Int. Conf. Comput. Vis., 1365--1374. IEEE

2019
[39]

Ulyanov, D.; Lebedev, V.; Lempitsky, V.; et al. 2016. Texture Networks: Feed-forward Synthesis of Textures and Stylized Images. In Proc. Int. Conf. Mach. Learn., 1349--1357. PMLR

2016
[40]

Wang, H.; Li, Y.; Wang, Y.; Hu, H.; and Yang, M.-H. 2020. Collaborative Distillation for Ultra-Resolution Universal Style Transfer. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 1860--1869. IEEE

2020
[41]

Wang, Z.; Zhao, L.; Zuo, Z.; Li, A.; Chen, H.; Xing, W.; and Lu, D. 2023. MicroAST: towards super-fast ultra-resolution arbitrary style transfer. In Proc. AAAI Conf. Artif. Intell., volume 37, 2742--2750

2023
[42]

Wei, L.; Zhang, S.; Gao, W.; and Tian, Q. 2018. Person Transfer GAN to Bridge Domain Gap for Person Re-Identification. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 79--88. IEEE

2018
[43]

Wu, G.; and Gong, S. 2021. Generalising without Forgetting for Lifelong Person Re-identification. In Proc. AAAI Conf. Artif. Intell., 2889--2897

2021
[44]

Xiao, T.; Li, S.; Wang, B.; Lin, L.; and Wang, X. 2016. End-to-End Deep Learning for Person Search. arXiv:1604.01850

2016 arXiv
[45]

Xu, K.; Zhang, H.; Li, Y.; Peng, Y.; and Zhou, J. 2024 a . Mitigate Catastrophic Remembering via Continual Knowledge Purification for Noisy Lifelong Person Re-Identification. In ACM Multimedia 2024

2024
[46]

Xu, K.; Zou, X.; Peng, Y.; and Zhou, J. 2024 b . Distribution-Aware Knowledge Prototyping for Non-Exemplar Lifelong Person Re-Identification. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 16604--16613. IEEE

2024
[47]

Xu, K.; Zou, X.; and Zhou, J. 2024. LSTKC: Long Short-Term Knowledge Consolidation for Lifelong Person Re-identification. In Proc. AAAI Conf. Artif. Intell., volume 38, 16202--16210

2024
[48]

Yang, Z.; Wu, D.; Zhang, W.; Li, B.; and Wang, W. 2023. Handling Label Uncertainty for Camera Incremental Person Re-Identification. In Proc. ACM Int. Conf. Multimedia, 6253–6263. ISBN 9798400701085

2023
[49]

Yin, X.; Shi, J.; Zhang, Y.; Lu, Y.; Zhang, Z.; Xie, Y.; and Qu, Y. 2024. Robust pseudo-label learning with neighbor relation for unsupervised visible-infrared person re-identification. In Proc. ACM Int. Conf. Multimedia, 2242--2251

2024
[50]

Yu, C.; Shi, Y.; Liu, Z.; Gao, S.; and Wang, J. 2023. Lifelong Person Re-Identification via Knowledge Refreshing and Consolidation. In Proc. AAAI Conf. Artif. Intell., 3295--3303

2023
[51]

Zhao, B.; Tang, S.; Chen, D.; Bilen, H.; and Zhao, R. 2021 a . Continual Representation Learning for Biometric Identification. In Proc. IEEE Winter Conf. Appl. Comput. Vis, 1197--1207. IEEE

2021
[52]

Zhao, H.; Tian, M.; Sun, S.; Shao, J.; Yan, J.; Yi, S.; Wang, X.; and Tang, X. 2017. Spindle Net: Person Re-Identification with Human Body Region Guided Feature Decomposition and Fusion. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 907--915. IEEE

2017
[53]

Zhao, Y.; Zhong, Z.; Yang, F.; Luo, Z.; Lin, Y.; Li, S.; and Sebe, N. 2021 b . Learning to Generalize Unseen Domains via Memory-Based Multi-Source Meta-Learning for Person Re-Identification. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 6277--6286. IEEE

2021
[54]

Zheng, K.; Lan, C.; Zeng, W.; Zhang, Z.; and Zha, Z.-J. 2021. Exploiting Sample Uncertainty for Domain Adaptive Person Re-identification. In Proc. AAAI Conf. Artif. Intell., volume 35, 3538--3546

2021
[55]

Zheng, L.; Shen, L.; Tian, L.; Wang, S.; Wang, J.; and Tian, Q. 2015. Scalable Person Re-Identification: A Benchmark. In Proc. IEEE Int. Conf. Comput. Vis., 1116--1124. IEEE

2015
[56]

Zhou, J.; Su, B.; and Wu, Y. 2022. Discriminative self-paced group-metric adaptation for online visual identification. IEEE Trans. Pattern Anal. Mach. Intell., 45(4): 4368--4383

2022
[57]

Zhou, J.; Yu, P.; Tang, W.; and Wu, Y. 2017. Efficient online local metric adaptation via negative samples for person re-identification. In Proc. IEEE Int. Conf. Comput. Vis., 2420--2428

2017
[58]

Zhou, K.; Yang, Y.; Qiao, Y.; and Xiang, T. 2024. Mixstyle Neural Networks for Domain Generalization and Adaptation. Int. J. Comput. Vis., 132(3): 822--836

2024
[59]

V.; and Kautz, J

Zou, Y.; Yang, X.; Yu, Z.; Kumar, B. V.; and Kautz, J. 2020. Joint Disentangling and Adaptation for Cross-Domain Person Re-Identification. In Proc. Eur. Conf. Comput. Vis., 87--104. Springer

2020

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

, " * write output.state after.block = add.period write newline

ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

[2] [2]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

[3] [3]

Branch, H. O. S. D. 2006. Imagery Library for Intelligent Detection Systems (i-lids). In 2006 IET Conf. on Crime and Security, 445--448. IET

work page 2006

[4] [4]

Chen, H.; Lagadec, B.; and Bremond, F. 2022. Unsupervised Lifelong Person Re-identification via Contrastive Rehearsal. arXiv:2203.06468

arXiv 2022

[5] [5]

Cui, Z.; Zhou, J.; Wang, X.; Zhu, M.; and Peng, Y. 2024. Learning Continual Compatible Representation for Re-indexing Free Lifelong Person Re-identification. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 16614--16623. IEEE

work page 2024

[6] [6]

Deng, W.; Zheng, L.; Ye, Q.; Kang, G.; Yang, Y.; and Jiao, J. 2018. Image-Image Domain Adaptation with Preserved Self-Similarity and Domain-Dissimilarity for Person Re-Identification. In Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 994--1003. IEEE

work page 2018

[7] [7]

Ge, W.; Du, J.; Wu, A.; Xian, Y.; Yan, K.; Huang, F.; and Zheng, W.-S. 2022. Lifelong Person Re-identification by Pseudo Task Knowledge Preservation. In Proc. AAAI Conf. Artif. Intell., 688--696

work page 2022

[8] [8]

Gray, D.; and Tao, H. 2008. Viewpoint Invariant Pedestrian Recognition with an Ensemble of Localized Features. In Proc. Eur. Conf. Comput. Vis., 262--275. Springer

work page 2008