DARA: Degradation-Aware Low-Rank Residual Adaptation with Original-to-Corrupted Distillation for Corruption-Robust Animal Re-Identification

Cynthia Xie; Talia Xu

REVIEW 3 major objections 4 minor 38 references

DARA: Degradation-Aware Low-Rank Residual Adaptation with Original-to-Corrupted Distillation for Corruption-Robust Animal Re-Identification

T0 review · 3 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read DARA repairs corrupted-query embeddings by routing low-rank residual experts on a frozen backbone, recovering 77% of the full corrupted fine-tuning gap with 0.49% added parameters.

desk verdict A practical, solid corruption-robust Re-ID retrofit whose headline generalization claim is under-supported because the unseen corruption set is never enumerated. read the letter →

arxiv 2607.16644 v1 pith:EZ26LZ7N submitted 2026-07-18 cs.CV

Cynthia Xie , Talia Xu This is my paper · ORCID

classification cs.CV

keywords animalre-identificationcorruptionrobustnesslow-rankadaptationmixture-of-expertsroutingknowledgedistillationfeature-spacerepairparameter-efficientfine-tuningretrieval

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

Animal re-identification models fail when query images are blurred, noisy, or compressed, because degradation shifts embeddings away from the original-image retrieval space. The paper argues this can be fixed without retraining the backbone: DARA freezes a standard fine-tuned model, attaches a small bank of low-rank residual experts, and uses a router that composes them per input using image statistics and feature cues. An original-image teacher anchors each repaired embedding and, more importantly, supervises the positive and negative pairs whose similarities corruption distorts most. On four animal datasets, the method beats standard and augmentation-based fine-tuning on corrupted-query retrieval, keeps unchanged-image performance close to the original model, and generalizes to unseen corruptions and a cross-domain setting. If sustained, this would mean most of the benefit of full corrupted fine-tuning is available as a lightweight, label-free retrofit.

What carries the argument

The degradation-aware router with a bank of three low-rank residual experts, inserted at stages 4 and 5 of a frozen compact backbone. Each expert is a low-rank factorized channel-wise correction (down-projection then up-projection), and the router selects and soft-weight-combines the top two experts per input using a 25-dimensional cue: nine hand-crafted corruption statistics (sharpness, exposure, frequency content, compression artifacts) concatenated with a 16-dimensional projected feature descriptor. This mechanism makes the repair input-conditioned without corruption labels; original-to-corrupted distillation (individual embedding alignment plus relation-level alignment on the hardest tea

What would settle it

Run DARA on a set of naturally degraded query images (e.g., motion-blurred camera-trap photos, low-light sensor noise, and mixed JPEG compression) and compare corrupted-query mAP to standard fine-tuning. If the retrofit shows no improvement on these natural degradations, or fails on any of the eight synthetic corruption types outside its training set, the claim of generalization beyond synthetic corruptions would be refuted.

Watch

Extended reading notes

Core claim

The central claim is that degradation shifts in animal Re-ID embeddings can be repaired in feature space rather than by pixel restoration or full retraining. A router with a soft top-2 mechanism composes three low-rank residual experts inserted at the last two feature stages; each expert predicts a channel-wise correction via a low-rank factorization, and the router weighs the top two selected experts from a 25-dimensional cue that combines nine hand-crafted image-quality statistics with a 16-dimensional learned feature descriptor. Because no corruption-type labels are used, expert specialization emerges from the retrieval objective. A frozen original-image teacher anchors each repaired embe

Load-bearing premise

The eight synthetic corruption operators used in training and evaluation are representative enough of real-world blur, noise, compression, and their mixtures that the router's hand-crafted statistics and learned expert corrections transfer to field conditions.

Editorial extensions

If this is right

Retrofitting a frozen backbone with DARA recovers 77% of the corrupted-query mAP gap to full corrupted fine-tuning, so most robustness is achievable without retraining.
The 0.49% parameter and 0.05% FLOP overhead make the retrofit suitable for resource-constrained deployments on edge devices.
Learned routing outperforms fixed expert mixtures, showing input-conditioned composition, not capacity alone, drives robustness.
DARA improves on unseen corruptions and on a cross-domain cattle dataset, suggesting the learned repairs are not overfit to the eight training corruption types.
Unchanged-image retrieval stays close to the original fine-tuned model, so the robustness gain does not come at the cost of standard Re-ID accuracy.

Reading between the lines

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

A natural extension is to replace the hand-crafted 9-dim degradation statistics with a learned degradation encoder; if real-world blur and noise deviate from synthetic operators, a learned cue may transfer better — but this goes beyond the paper, which defers natural degradations to future work.
The relation-level distillation that mines pairs with the largest teacher–student similarity gaps is a general recipe for retrieval tasks where corruption reorders neighbours; it could be applied to person Re-ID or product matching.
The router weights vary across corruption conditions, so the expert bank could double as a corruption-type indicator, providing interpretable degradation cues at inference without labels.
A direct evaluation on camera-trap images with natural motion blur and low-light sensor noise would test whether the synthetic-to-real transfer holds; the authors list this as future work.

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

You should know two things about this paper. First, the core method is a genuine composition of existing ideas: routed low-rank residual experts, a hand-crafted degradation cue plus learned feature cue, and original-to-corrupted distillation that mines the hardest retrieval relations. The pieces exist in cited work, but the combination is new and the empirical gains are real and consistent across four animal Re-ID datasets. Second, the claim that DARA 'generalizes to unseen corruptions' is currently unfalsifiable. The paper defines the eight training corruptions in §4.1.2, but Fig. 3 and Table 6 never say which held-out corruptions are used, nor at what severities. Without that list you cannot reproduce the experiment or tell whether the router is actually transferring to new degradation families or just seeing different levels of the same eight types. That is a load-bearing soft spot, not a cosmetic one.

What the paper does well: it frames corruption robustness as input-conditioned feature-space repair, which is a useful and clearly explained reframing. The frozen-backbone retrofit is well motivated for compact deployment models, and the efficiency numbers (0.49% parameters, 0.05% FLOPs, trivial latency increase) are plausible. The ablations are well designed and the component contributions are cleanly isolated: one adapted layer gives the largest jump, the second adds less, distillation helps most on harder matches, and learned top-2 routing beats fixed mixtures. The paper also honestly defers naturally occurring degradations to future work.

The soft spots, in proportion: the missing unseen-corruption specification is the main one. Also, the 10-trial averages come without standard deviations, the cross-domain experiment uses a single seed, key hyperparameters (lambda_ind, lambda_rel, r_l, hard-mining details) are not reported, and no code or data is released. These are all fixable and do not undermine the central result. The teacher is a frozen original-image backbone, and distilling toward original-image geometry is a stated objective, not a disguised test-metric reuse, so I do not see circularity.

This is a subfield-level contribution — practical, deployable, and likely useful for people working on corruption-robust animal Re-ID. It is not a field reorg. But it is a solid paper that deserves a serious referee. My recommendation: send it to peer review, but require the authors to enumerate the unseen corruption set, report variance across trials and seeds, and release code and hyperparameters before the generalization claim is accepted.

Referee Report

3 major / 4 minor

Summary. The paper introduces DARA, a lightweight retrofit for compact animal re-identification models that aims to make them robust to common image corruptions on query images. The method freezes the pretrained backbone and adds routed low-rank residual experts at two feature stages; a soft top-2 router combines experts using a hand-crafted 9-dimensional degradation statistic plus a 16-dimensional feature cue. Training uses original-to-corrupted distillation from a frozen original-image teacher, combining individual embedding alignment and hard-pair relation-level alignment with standard Re-ID losses. Experiments on ATRW, FriesianCattle2017, MPDD, and SeaStarReID2023 report that DARA improves corrupted-query mAP over standard and AugMix fine-tuning (55.70 vs 41.67 average), recovers 77.0% of the gap to full corrupted fine-tuning while adding 0.49% parameters and 0.05% FLOPs, and claims generalization to unseen corruptions and cross-domain transfer.

Significance. If substantiated, the result is practically useful: it demonstrates that corruption robustness for compact animal Re-ID can be obtained by repairing feature-space geometry with a small set of routed experts, rather than by full corrupted retraining. The paper is clearly written and supported by careful ablations of the expert pool, routing mechanism, distillation terms, and adapted stages. The central claim is internally consistent, and I see no circularity in the evaluation: the frozen teacher is used only as a distillation reference, not as a test-time oracle. However, two load-bearing aspects of the evaluation are currently underspecified or under-reported: the set of 'unseen' corruptions is never enumerated, and no variance information is given for the 10-trial protocol or the single-seed cross-domain experiment. These issues must be fixed before the generalization claims can be accepted.

major comments (3)

[§4.2.3 / Fig. 3 / Table 6] The held-out corruption set behind the 'unseen corruption' claim is never enumerated. Section 4.1.2 defines the eight training/evaluation corruptions, but neither §4.2.3, Fig. 3, nor Appendix Table 6 specifies which corruption operators or severity levels are used for the 'Unseen CQ' results. Without this, the third contribution ('generalizes to unseen corruptions') cannot be independently verified or reproduced, and the reader cannot tell whether the model is facing genuinely new degradation families or simply different severities of the same eight ImageNet-C types. Please specify the exact held-out operators, severity levels, and sampling protocol, and ideally report per-corruption breakdowns.
[§4.1.2 / Tables 1, 3, 4, 6] The protocol states that the random corruption draw is repeated for 10 trials and the average is reported, but no standard deviations, confidence intervals, or per-trial values are provided in any table. The headline comparison (DARA 55.70 vs AugMix 41.67 average CQ mAP) is large, but on individual datasets the margins are smaller (e.g., SeaStarReID2023 37.53 vs 30.34); without variance the reader cannot assess whether the gains are significant. Cross-domain transfer (§4.2.3, Table 8) is reported for a single seed (seed 45). Please report mean±std over the 10 trials and multiple seeds for the cross-domain experiment.
[§3.3 / §4.1.4] The optimization coefficients λ_ind and λ_rel and the low-rank dimension r_l are never specified, although the method description says λ_ind and λ_rel balance the distillation terms and r_l ≪ C_l. These are part of the method's definition, and without their values the experimental setup is not fully reproducible. Please include these values in the implementation details section.

minor comments (4)

[Table 7] The header structure of Table 7 is hard to parse: the columns 'Corrupted Eval', 'Corrupted Query', and 'Corrupted Gallery' seem to overlap or duplicate. Please restructure the header so each evaluation setting has clearly separated subcolumns.
[Fig. 2] The routing-weight heatmap has no color bar or numeric labels, which makes the claimed condition-dependent routing behavior difficult to quantify. Adding a color scale or a small table of average weights would help.
[§4.2.3 / Appendix A.4] The CornwallCattle dataset is referenced without any description (e.g., number of identities, images, or collection setting). A short sentence or a citation to a dataset paper would improve reproducibility.
[References] Several references are listed with 2026 publication dates and arXiv identifiers; please verify the final publication status and correct any preprint/venue inconsistencies (e.g., refs. [1], [2], [20], [23], [28], [36]).

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; DARA's main numbers are held-out empirical results, and the distillation loss is a training objective rather than the evaluation metric.

full rationale

DARA's central claims are empirical rather than derivational. The model is trained on synthetic corrupted views, and the reported corrupted-query mAP and 77.0% gap-recovery figures are computed on held-out corrupted queries under the protocol in Section 4.1.2 and reported in Tables 1-2. The original-to-corrupted distillation losses (Lind and Lrel) minimize a training proxy, matching corrupted-student embeddings and hard pair relations to original-image teacher geometry, which is the method's stated objective, not a disguised reuse of the test metric. At inference the teacher is removed and gallery embeddings are produced by the adapted student (Section 3.3), so the evaluation is not an identity of loss and metric. No parameter is fitted to the reported CQ mAP and then renamed as a prediction. The only self-citations are contextual or dataset-providing ([9] for CornwallCattle and resource-constrained motivation, [19] for a related multimodal dataset); they are not load-bearing justifications of the core mechanism. One genuine evidence gap is flagged but is not circularity: the unseen corruptions behind Section 4.2.3, Figure 3, and Appendix Table 6 are never enumerated, neither the operators nor severity levels, so the unseen-generalization claim is hard to verify and reproduce. That is a missing-support/reproducibility problem, not a reduction of the claim to its own inputs. Accordingly, the circularity score is minimal (1), reflecting only a minor non-load-bearing self-citation and no circular derivation.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

DARA is an empirical ML paper with no mathematical derivation and no new physical entities. The central claims rest on the standard deep-learning assumptions of the Re-ID backbone, on the validity of synthetic ImageNet-C corruption as a proxy for real degradations, and on several unreported hyperparameter choices (λ, r_l) that shape the method's capacity.

free parameters (4)

Distillation coefficients λ_ind, λ_rel = not reported
They control the strength of the teacher anchor in §3.3 but no numerical values are given.
Low-rank dimension r_l = not reported
The size of A_l/B_l matrices in §3.1 is left unspecified (r_l ≪ C_l only), directly setting expert capacity and parameter count.
Expert-pool/hard-mining hyperparameters = M=3, top-2, K=8
These are hand-chosen (§3.1–3.2); only pool size is partially ablated (M=4), and K is not varied.
Degradation-cue z-score statistics = not reported
The 9-dimensional statistics d(x) are z-score normalized in §3.1, presumably with training-set mean/std, but these values are not disclosed.

assumptions (5)

domain assumption Synthetic corruption preserves the identity label of each image.
§3.2 states 'Identity labels remain unchanged under corruption'; Re-ID loss and relation distillation assume the corrupted view matches the same gallery individual.
domain assumption Original-image teacher embeddings are the correct target geometry for repaired corrupted inputs.
The teacher is a frozen Std-FT model (§3.2); if aligning corrupted features to this geometry fabricates identity cues rather than recovering them, the central retrieval gain would not transfer to real degradations.
domain assumption ImageNet-C eight operators at severities 1–5 are a representative proxy for field degradations.
Training and evaluation corruptions are sampled from these operators (§4.1.2), and the conclusion explicitly defers natural degradations to future work.
ad hoc to paper The hand-crafted degradation statistics plus stage-3 feature cue are sufficient for the router to generalize to unseen corruptions.
The 25-dim router cue (§3.1) is designed without a proof of sufficiency; unseen-corruption gains depend on this premise.
ad hoc to paper Channel-wise low-rank residuals broadcast spatially can correct corruption-induced feature shifts.
Residuals are computed from pooled channel descriptors and added uniformly across spatial positions (§3.1); spatially localized degradations may not be fully repairable by this form.

how reviews work

0 comments

Cite this review

Pith. "Pith review of DARA: Degradation-Aware Low-Rank Residual Adaptation with Original-to-Corrupted Distillation for Corruption-Robust Animal Re-Identification." pith.science (2026). https://pith.science/paper/EZ26LZ7N

@misc{pith2026260716644,
  author       = {Pith},
  title        = {Pith review of: DARA: Degradation-Aware Low-Rank Residual Adaptation with Original-to-Corrupted Distillation for Corruption-Robust Animal Re-Identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EZ26LZ7N}},
  note         = {Machine review of arXiv:2607.16644}
}

read the original abstract

Animal re-identification (Re-ID) relies on fine-grained identity cues that can be disrupted by blur, noise, compression, and other visual degradations. Existing robustness strategies based on degradation-augmented training or pixel-level restoration improve robustness indirectly, but do not explicitly repair shifts in the identity retrieval space. We study corruption-robust animal Re-ID as input-conditioned feature-space repair and introduce DARA, a lightweight retrofit for compact Re-ID models. DARA freezes the fine-tuned backbone and learns routed low-rank residual experts to adapt degraded-input embeddings without corruption-type annotations. To stabilize this adaptive repair, original-to-corrupted distillation uses an original-image teacher to preserve individual embeddings and retrieval relations. Experiments on ATRW, FriesianCattle2017, MPDD, and SeaStarReID2023 show that DARA improves corrupted-query retrieval over standard and augmentation-based fine-tuning, generalizes to unseen corruptions and cross-domain evaluation, and recovers 77.0% of the corrupted-query mAP gap to full corrupted fine-tuning while adding only 0.49% parameters and 0.05% FLOPs.

Figures

Figures reproduced from arXiv: 2607.16644 by the authors.

**Figure 1.** Overall framework of DARA for corruption-robust animal Re-ID. During training, each original image is paired with a corrupted [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

**Figure 3.** Unseen corrupted-query evaluation. (a) Absolute mAP [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 1 linked inside Pith

[1]

Wildlifereid-10k: Wildlife re-identification dataset with 10k individual animals

Luk ´aˇs Adam, V ojt ˇech ˇCerm´ak, Lukas Picek, and Kostas Papafitsoros. Wildlifereid-10k: Wildlife re-identification dataset with 10k individual animals. In2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 2090–2100, 2025. 2
[2]

Overview of animalclef 2025: Recognizing individual animals in images

Luk ´as Adam, Kostas Papafitsoros, Roman Kov ´ar, V ojtech Cerm´ak, and Luk ´as Picek. Overview of animalclef 2025: Recognizing individual animals in images. InWorking Notes of the Conference and Labs of the Evaluation Forum, CLEF 2025, Madrid, Spain, 9-12 September 2025, pages 2898–

2025
[3]

Visual localisation and individual identification of holstein friesian cattle via deep learning

William Andrew, Colin Greatwood, and Tilo Burghardt. Visual localisation and individual identification of holstein friesian cattle via deep learning. InProceedings of the IEEE International Conference on Computer Vision Workshops,
[4]

From species-specific models to universal re-id: a survey of animal re-identification.Information Fusion, 133:104323, 2026

Cigdem Beyan, Anil Osman Tur, and Ehsan Karimi. From species-specific models to universal re-id: a survey of animal re-identification.Information Fusion, 133:104323, 2026. 1, 2

2026
[5]

Dy- namic adapter with semantics disentangling for cross-lingual cross-modal retrieval, 2024

Rui Cai, Zhiyu Dong, Jianfeng Dong, and Xun Wang. Dy- namic adapter with semantics disentangling for cross-lingual cross-modal retrieval, 2024. 1, 3

2024
[6]

Wildlifedatasets: An open-source toolkit for ani- mal re-identification

V ojtˇech ˇCerm´ak, Lukas Picek, Luk´aˇs Adam, and Kostas Pa- pafitsoros. Wildlifedatasets: An open-source toolkit for ani- mal re-identification. InProceedings of the IEEE/CVF Win- ter Conference on Applications of Computer Vision (WACV),
[7]

Unire- store: Unified perceptual and task-oriented image restoration model using diffusion prior, 2025

I-Hsiang Chen, Wei-Ting Chen, Yu-Wei Liu, Yuan-Chun Chiang, Sy-Yen Kuo, and Ming-Hsuan Yang. Unire- store: Unified perceptual and task-oriented image restoration model using diffusion prior, 2025. 1, 2

2025
[8]

Bench- marks for corruption invariant person re-identification, 2022

Minghui Chen, Zhiqiang Wang, and Feng Zheng. Bench- marks for corruption invariant person re-identification, 2022. 2, 6

2022

Show all 38 references

[9]

Animal re-identification on microcontrollers, 2025

Yubo Chen, Di Zhao, Yun Sing Koh, and Talia Xu. Animal re-identification on microcontrollers, 2025. 1, 8

2025
[10]

LoRA+: Effi- cient low rank adaptation of large models

Soufiane Hayou, Nikhil Ghosh, and Bin Yu. LoRA+: Effi- cient low rank adaptation of large models. InProceedings of the 41st International Conference on Machine Learning, pages 17783–17806. PMLR, 2024. 3

2024
[11]

Animal re-identification algorithm for posture diversity

Zhimin He, Jiangbo Qian, Diqun Yan, Chong Wang, and Yu Xin. Animal re-identification algorithm for posture diversity. InICASSP 2023 – 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1–5, 2023. 5

2023
[12]

Benchmarking neu- ral network robustness to common corruptions and perturba- tions

Dan Hendrycks and Thomas Dietterich. Benchmarking neu- ral network robustness to common corruptions and perturba- tions. InInternational Conference on Learning Representa- tions, 2019. 1, 6

2019
[13]

Distilling the knowledge in a neural network

Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the knowledge in a neural network. InNeurIPS Deep Learning and Representation Learning Workshop, 2015. 3

2015
[14]

Openanimals: Re- visiting person re-identification for animals towards bet- ter generalization

Saihui Hou, Panjian Huang, Zengbin Wang, Yuan Liu, Zeyu Li, Man Zhang, and Yongzhen Huang. Openanimals: Re- visiting person re-identification for animals towards bet- ter generalization. InProceedings of the IEEE/CVF In- ternational Conference on Computer Vision (ICCV), pages ...

2025
[15]

Parameter-efficient transfer learning for nlp

Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin De Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. Parameter-efficient transfer learning for nlp. InProceedings of the 36th International Conference on Machine Learning, pages 2790–2799. PMLR,
[16]

Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-rank adaptation of large language models. InIn- ternational Conference on Learning Representations, 2022. 1, 3

2022
[17]

Real-world person re-identification via degra- dation invariance learning

Yukun Huang, Zheng-Jun Zha, Xueyang Fu, Richang Hong, and Liang Li. Real-world person re-identification via degra- dation invariance learning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 14072–14082, Seattle, W A, USA, 2020...

2020
[18]

Atrw: A benchmark for amur tiger re-identification in the wild

Shuyuan Li, Jianguo Li, Hanlin Tang, Rui Qian, and Weiyao Lin. Atrw: A benchmark for amur tiger re-identification in the wild. InProceedings of the IEEE/CVF International Conference on Computer Vision Workshops, 2019. 5

2019
[19]

Metawild: A multimodal dataset for animal re-identification with environmental metadata

Yuzhuo Li, Di Zhao, Tingrui Qiao, Yihao Wu, Bo Pang, and Yun Sing Koh. Metawild: A multimodal dataset for animal re-identification with environmental metadata. InProceed- ings of the 33rd ACM International Conference on Multime- dia, 2025. 2

2025
[20]

Parameters as ex- perts: Adapting vision models with dynamic parameter rout- ing, 2026

Meng Lou, Stanley Yu, and Yizhou Yu. Parameters as ex- perts: Adapting vision models with dynamic parameter rout- ing, 2026. 1, 3

2026
[21]

Low-rank expert merging for multi-source domain adaptation in person re-identification

Taha Mustapha Nehdi, Nairouz Mrabah, Atif Belal, Marco Pedersoli, and Eric Granger. Low-rank expert merging for multi-source domain adaptation in person re-identification. InProceedings of the IEEE/CVF Winter Conference on Ap- plications of Computer Vision (WACV), 2026. 3

2026
[22]

Rela- tional knowledge distillation, 2019

Wonpyo Park, Dongju Kim, Yan Lu, and Minsu Cho. Rela- tional knowledge distillation, 2019. 3

2019
[23]

Degradation-based augmented training for robust individual animal re-identification, 2026

Thanos Polychronou, Luk ´aˇs Adam, Viktor Penchev, and Kostas Papafitsoros. Degradation-based augmented training for robust individual animal re-identification, 2026. 1, 2

2026
[24]

Robust re-identification by multiple views knowledge distil- lation, 2020

Angelo Porrello, Luca Bergamini, and Simone Calderara. Robust re-identification by multiple views knowledge distil- lation, 2020. 3

2020
[25]

Auto-selected knowledge adapters for lifelong person re- identification.arXiv preprint arXiv:2405.19005, 2024

Xuelin Qian, Ruiqi Wu, Gong Cheng, and Junwei Han. Auto-selected knowledge adapters for lifelong person re- identification.arXiv preprint arXiv:2405.19005, 2024. 3

2024 arXiv
[26]

Ravoor and T

Prashanth C. Ravoor and T. S. B. Sudarshan. Deep learning methods for multi-species animal re-identification and track- ing – a survey.Comput. Sci. Rev., 38(C), 2020. 1, 2

2020
[27]

MobileNetV2: Inverted residuals and linear bottlenecks

Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zh- moginov, and Liang-Chieh Chen. MobileNetV2: Inverted residuals and linear bottlenecks. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 4510–4520, 2018. 6

2018
[28]

RobuMTL: Enhancing multi-task learning robustness against weather conditions

Tasneem Shaffee and Sherief Reda. RobuMTL: Enhancing multi-task learning robustness against weather conditions. In Proceedings of the IEEE/CVF Winter Conference on Appli- cations of Computer Vision (WACV), 2026. 3

2026
[29]

Wahltinez

Oscar Wahltinez and Sarah J. Wahltinez. An open-source general purpose machine learning framework for individual animal re-identification using few-shot learning.Methods in Ecology and Evolution, 15(2):373–387, 2024. 6

2024
[30]

Distilled person re-identification: Towards a more scal- able system

Ancong Wu, Wei-Shi Zheng, Xiaowei Guo, and Jian-Huang Lai. Distilled person re-identification: Towards a more scal- able system. In2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 1187–1196,
[31]

Glen, Jingfeng Zhang, Gillian Dobbie, Daniel Wilson, and Yun Sing Koh

Yihao Wu, Di Zhao, Yuzhuo Li, Matthew Alajas, Alistair S. Glen, Jingfeng Zhang, Gillian Dobbie, Daniel Wilson, and Yun Sing Koh. Overcoming fine-grained visual challenges in animal re-identification via semantic feature alignment. In Proceedings of the IEEE/CVF Winter Conferen...

2026
[32]

Parameter-efficient fine-tuning for pre-trained vision models: A survey, 2025

Yi Xin, Jianjiang Yang, Siqi Luo, Yuntao Du, Qi Qin, Kan- grui Cen, Yangfan He, Zhiwei Zhang, Bin Fu, Xiaokang Yang, Guangtao Zhai, Ming-Hsuan Yang, and Xiaohong Liu. Parameter-efficient fine-tuning for pre-trained vision models: A survey, 2025. 3

2025
[33]

Visual recognition-driven image restoration for multiple degradation with intrinsic se- mantics recovery

Zizheng Yang, Jie Huang, Jiahao Chang, Man Zhou, Hu Yu, Jinghao Zhang, and Feng Zhao. Visual recognition-driven image restoration for multiple degradation with intrinsic se- mantics recovery. In2023 IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR), pages...

2023
[34]

Addressing the elephant in the room: Robust animal re-identification with unsupervised part-based feature alignment, 2024

Yingxue Yu, Vidit Vidit, Andrey Davydov, Martin En- gilberge, and Pascal Fua. Addressing the elephant in the room: Robust animal re-identification with unsupervised part-based feature alignment, 2024. 2

2024
[35]

Corruption- invariant person re-identification via coarse-to-fine feature alignment.IEEE Transactions on Circuits and Systems for Video Technology, 35(2):1084–1097, 2025

Xinyu Zhang, Peng Zhang, and Caifeng Shan. Corruption- invariant person re-identification via coarse-to-fine feature alignment.IEEE Transactions on Circuits and Systems for Video Technology, 35(2):1084–1097, 2025. 3

2025
[36]

Robust mixed- degradation person re-identification via structural consis- tency distillation.Pattern Recognition, 179:113938, 2026

Siyuan Zhao, Wenxin Huang, Wenxuan Liu, Xuemei Jia, Xiyu Han, Siqi Liang, and Xian Zhong. Robust mixed- degradation person re-identification via structural consis- tency distillation.Pattern Recognition, 179:113938, 2026. 3

2026
[37]

Pose-guided feature learning with knowledge distillation for occluded person re- identification, 2021

Kecheng Zheng, Cuiling Lan, Wenjun Zeng, Jiawei Liu, Zhizheng Zhang, and Zheng-Jun Zha. Pose-guided feature learning with knowledge distillation for occluded person re- identification, 2021. 3 A. Additional Experimental Results A.1. Dataset Statistics Table 5. Statistics of th...

2021
[2908]

CEUR-WS.org, 2025. 2

2025

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.

[1] [1]

Wildlifereid-10k: Wildlife re-identification dataset with 10k individual animals

Luk ´aˇs Adam, V ojt ˇech ˇCerm´ak, Lukas Picek, and Kostas Papafitsoros. Wildlifereid-10k: Wildlife re-identification dataset with 10k individual animals. In2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 2090–2100, 2025. 2

[2] [2]

Overview of animalclef 2025: Recognizing individual animals in images

Luk ´as Adam, Kostas Papafitsoros, Roman Kov ´ar, V ojtech Cerm´ak, and Luk ´as Picek. Overview of animalclef 2025: Recognizing individual animals in images. InWorking Notes of the Conference and Labs of the Evaluation Forum, CLEF 2025, Madrid, Spain, 9-12 September 2025, pages 2898–

2025

[3] [3]

Visual localisation and individual identification of holstein friesian cattle via deep learning

William Andrew, Colin Greatwood, and Tilo Burghardt. Visual localisation and individual identification of holstein friesian cattle via deep learning. InProceedings of the IEEE International Conference on Computer Vision Workshops,

[4] [4]

From species-specific models to universal re-id: a survey of animal re-identification.Information Fusion, 133:104323, 2026

Cigdem Beyan, Anil Osman Tur, and Ehsan Karimi. From species-specific models to universal re-id: a survey of animal re-identification.Information Fusion, 133:104323, 2026. 1, 2

2026

[5] [5]

Dy- namic adapter with semantics disentangling for cross-lingual cross-modal retrieval, 2024

Rui Cai, Zhiyu Dong, Jianfeng Dong, and Xun Wang. Dy- namic adapter with semantics disentangling for cross-lingual cross-modal retrieval, 2024. 1, 3

2024

[6] [6]

Wildlifedatasets: An open-source toolkit for ani- mal re-identification

V ojtˇech ˇCerm´ak, Lukas Picek, Luk´aˇs Adam, and Kostas Pa- pafitsoros. Wildlifedatasets: An open-source toolkit for ani- mal re-identification. InProceedings of the IEEE/CVF Win- ter Conference on Applications of Computer Vision (WACV),

[7] [7]

Unire- store: Unified perceptual and task-oriented image restoration model using diffusion prior, 2025

I-Hsiang Chen, Wei-Ting Chen, Yu-Wei Liu, Yuan-Chun Chiang, Sy-Yen Kuo, and Ming-Hsuan Yang. Unire- store: Unified perceptual and task-oriented image restoration model using diffusion prior, 2025. 1, 2

2025

[8] [8]

Bench- marks for corruption invariant person re-identification, 2022

Minghui Chen, Zhiqiang Wang, and Feng Zheng. Bench- marks for corruption invariant person re-identification, 2022. 2, 6

2022