Pith. sign in

REVIEW 4 major objections 5 minor 71 references

SapiensID 2.0: Aligning Human Recognition Foundation Models with Human Perception

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read SapiensID 2.0 claims that a single model can recognize faces, re-identify people across clothing changes, and read gait from video, by aligning its embedding space with invariant soft biometrics and tracking keypoint motion over time.

desk verdict Plausible new components (ITA/TND/K-SAH) and strong reported zero-shot gains, but the headline SOTA rests on an unverified disjointness assumption between WebBody4M and the test sets. read the letter →

arxiv 2608.10497 v1 pith:VCRA5REW submitted 2026-08-11 cs.CV

Yiyang Su , Jie Zhu , Feng Liu , Anil K. Jain , Xiaoming Liu This is my paper · ORCID
classification cs.CV
keywords personre-identificationgaitrecognitionsoftbiometricssemanticdistillationmultimodallargelanguagemodelskinematicattentionclothing-changeface
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 argues that current human recognition models anchor identity to static image texture, and that this causes 'semantic blindness': they overweigh transient traits such as clothing while ignoring stable soft biometrics, and they discard the motion signatures that identify a person by gait. SapiensID 2.0 is proposed as a single foundation model that injects both missing cues into one embedding space. Semantic knowledge is distilled offline from a multimodal large language model: Invariant Trait Alignment (ITA) pulls the student representation toward the principal components of fixed traits such as gender and body build, and Transient Noise Disentanglement (TND) pushes it orthogonal to the principal components of clothing and other transient attributes. Temporal knowledge is added by a Kinematic Semantic Attention Head (K-SAH) that tracks keypoint-anchored semantic patches across a short temporal window, so an image-trained model can exploit gait without video retraining. Because the teacher model and projection layers are discarded after training, the paper claims all of these gains come at zero additional inference cost and without degrading face recognition.

What carries the argument

The load-bearing machinery is a bipartite semantic subspace alignment plus a temporal attention head called K-SAH. ITA uses SVD of the teacher's fixed-trait text embeddings to form a basis $V_T^F$ and aligns the student's projected feature principal components to it, combining a cosine-similarity term on the basis vectors with a mean-squared-error term on the projected coordinates. TND builds a basis $V_T^D$ from the dynamic-trait embeddings and adds the penalty $\lVert Y V_T^D \rVert_F^2$, which projects the student feature out of the transient-noise subspace. K-SAH extends the spatial semantic attention across a temporal window: it repeats the pretrained 2D positional embedding over time, lets the 17 keypoint-anchored queries attend to the concatenated features of neighboring frames, and pools with an occlusion mask, at cost linear in the window size. Because the teacher and projection MLP are dropped after training, inference uses only the 63M-parameter ViT backbone.

What would settle it

An identity-overlap audit between WebBody4M and each test set, for example by clustering face or body embeddings from both sides and checking whether any same-identity pairs cross the train/test boundary, would settle the zero-shot claim; a second, cleaner test is evaluating on a newly collected benchmark that is guaranteed disjoint and seeing whether the reported gains persist.

Watch

Extended reading notes

Core claim

The paper's central claim is that human recognition is fundamentally about invariant anatomical traits and continuous motion, and that a foundation model can be made to use both by aligning its embedding space with semantic subspaces derived from a vision-language teacher. ITA performs this alignment for fixed soft biometrics by computing the top $D_F=6$ principal components of the teacher's trait embeddings (gender, race, age, skin tone, body build, facial structure) and matching the student's projected principal components to them. TND handles dynamic soft biometrics by extracting the top $D_D=5$ principal components of hair, facial hair, accessories, and clothing embeddings and penalizing the student's projection onto that subspace, forcing orthogonality to transient appearance. K-SAH replaces per-frame spatial pooling with cross-frame attention along the 17 keypoint-anchored semantic queries, tracking how those patches move across a temporal window and pooling through an occlusion mask. The paper reports that this single image-trained model outperforms specialized baselines on video re-identification and gait recognition, improves clothing-change re-identification, and preserves or slightly improves face verification.

Load-bearing premise

The reported zero-shot results assume the WebBody4M training corpus contains no identities that also appear in the test benchmarks; if the training set overlaps with any test set, the state-of-the-art numbers are inflated.

Editorial extensions

If this is right

  • Clothing-change person re-identification should improve substantially, since the model is explicitly prohibited from using clothing variance; on LTCC and PRCC clothing-change protocols the paper reports large gains in both rank-1 and mAP over its SapiensID baseline.
  • A model trained almost entirely on static images can still recognize gait if its attention tracks keypoint-anchored patches across frames; the paper reports raising CCPG clothing-change rank-1 from 33.9% to 60.8% with a five-frame window.
  • Face recognition need not be sacrificed when body-level semantic alignment is added; five face benchmarks show accuracy at or slightly above the baseline, which the paper attributes to ITA's invariant traits correlating with facial structure.
  • The semantic and kinematic gains carry no extra inference cost, because the MLLM teacher and its projection layers are discarded; inference remains a single forward pass on the 63M-parameter backbone.

Reading between the lines

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

  • If the alignment recipe transfers, the same ITA/TND structure could be applied to other vision foundation models where transient appearance contaminates identity, such as animal or vehicle re-identification; the paper does not test these domains.
  • Because K-SAH's cost is linear in the temporal window and the static head is its $T=1$ special case, full tracklet inference without retraining is a plausible extension; a direct check would compare five-frame windows against longer ones on CCPG or CCVID.
  • The teacher's zero-shot labels are the only semantic supervision, so swapping the vision-language teacher and re-running the distillation should shift student behavior in predictable ways; that experiment would separate the distillation mechanism's contribution from the teacher's specific knowledge.
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

4 major / 5 minor

Summary. The paper proposes SapiensID 2.0, a unified human recognition foundation model that extends the earlier SapiensID model with explicit semantic awareness and temporal kinematic modeling. Semantic awareness is injected through offline distillation from a frozen MLLM: Invariant Trait Alignment (ITA) aligns the student embedding with principal components of fixed soft-biometric attributes, and Transient Noise Disentanglement (TND) penalizes projection onto dynamic attributes such as clothing. Temporal modeling is handled by a Kinematic Semantic Attention Head (K-SAH) that extends the spatial semantic attention over a local temporal window while preserving the static architecture as the T=1 special case. The model is trained once on WebBody4M and evaluated zero-shot on video and image person re-ID, gait recognition, and face verification benchmarks. The authors report state-of-the-art results on CCVID, CCPG, LTCC, PRCC, and several face benchmarks, with zero additional inference overhead.

Significance. If the reported results hold up, the contribution is significant: the paper demonstrates that a largely image-pretrained backbone can be lifted to video-based tasks without video-scale retraining, and that MLLM-derived semantic subspaces can improve clothing-change robustness in a unified biometric model. The method is clearly motivated, the losses in Eqs. (3)-(5) and Eq. (8) are mathematically well defined, and the ablations in Table 4 are internally consistent. The zero-inference-overhead property is an important practical strength. However, the central zero-shot SOTA claim rests on two premises that are not yet established: identity disjointness between WebBody4M and the test sets, and a genuinely fair zero-shot comparison against the baselines. These points are load-bearing and must be addressed before the claims can be accepted.

major comments (4)
  1. [Section 4 and A.1] The zero-shot claim requires that identities in WebBody4M are disjoint from identities in every test set, but no identity-overlap analysis is reported for CCVID, CCPG, LTCC, PRCC, Market1501, MSMT17, CCDA, Celeb-ReID, LFW, CPLFW, CALFW, AgeDB, or CFP-FP. This matters because WebBody4M is described as collected via keyword queries and face similarities (Section 1), and several test sets are web- or celebrity-derived. Please provide per-dataset overlap statistics or explain why overlap is impossible by construction; without this, the headline zero-shot SOTA could be inflated by training-set memorization.
  2. [Section 3.4 and A.1] K-SAH is trained with temporal windows of T=5 frames, but the paper states that a single model is trained strictly on WebBody4M (Section 4) and that WebBody4M consists of more than 90% static images (Section 1). The construction of the video tracklets used to train K-SAH is never described: how many tracklets exist, how they are sampled, whether static images are used as T=1 inputs, and what augmentation is applied. This is a critical reproducibility gap because the CCPG and CCVID gains attributed to K-SAH depend on temporal training data that is not specified.
  3. [Section 4 vs. Table 6] The text claims that 'all models are evaluated zero-shot' for a fair comparison, but Table 6 explicitly lists baselines trained on target-domain splits: SOLIDER on LUP4M+Market1501 and LUP4M+MSMT17, and HAP on LUP4M+LTCC, LUP4M+PRCC, LUP4M+Market1501, and LUP4M+MSMT17. These baselines have seen test-domain training data, so the comparison is not zero-shot and the fairness statement in Section 4 is inaccurate. Please either retrain all baselines on the same non-target data or clearly label target-domain pretraining and rephrase the zero-shot claim to match the actual protocol.
  4. [Tables 1-3 and overall experiments] No error bars, confidence intervals, or multiple-seed results are reported. Several differences are small enough to be within typical training variance, for example Table 2a shows SapiensID 2.0 mAP on Market1501 at 67.15 versus SapiensID at 67.25, and Table 1 shows CCVID (CC) top-1 at 88.62 versus 88.72. Please report standard deviations over at least three runs, or otherwise justify that single-run numbers support the SOTA claims at the stated precision.
minor comments (5)
  1. [Table 5] The dataset name 'CelevReID' in Table 5 is a typo; it should be 'Celeb-ReID' to match the text and reference [23].
  2. [Equation (3)] The projection weight λ_P appears in Eq. (3) without definition; the value 0.5 is mentioned only later in Section A.1, so please introduce the symbol where the loss is first defined.
  3. [Table 1] On the CCVID (CC) protocol, SapiensID 2.0 slightly decreases top-1 accuracy (88.62 vs. 88.72) while improving mAP (78.04 vs. 72.22); the text highlights only the mAP gain and should acknowledge the top-1 trade-off.
  4. [Section 4] The abbreviation 'CL' in the CCPG evaluation is used without expansion; please define it as the clothing-changing protocol at first use, consistent with the CC terminology used elsewhere.
  5. [Section A.2] The sentence in Section A.2 that repeats 'one single model to test the zero-shot generalization capabilities' should be updated if the baseline metadata in Table 6 is corrected, since the target-domain pretraining of some baselines is inconsistent with a literal zero-shot reading.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the semantic and kinematic modules are trained via MLLM distillation on WebBody4M and evaluated on external benchmarks; self-citations supply only the baseline.

full rationale

SapiensID 2.0's derivation chain is not circular. The MLLM teacher (Qwen2.5-VL-7B) generates soft-biometric pseudo-labels on WebBody4M; ITA/TND losses align the student's projected embedding to the fixed subspace and orthogonalize the dynamic subspace; K-SAH extends spatial attention across temporal windows. Each component is then measured by retrieval and verification accuracy on external benchmarks (CCVID, CCPG, LTCC, PRCC, Market1501, MSMT17, CCDA, Celeb-ReID, LFW, CPLFW, CFP-FP, CALFW, AgeDB) and by ablations against the SapiensID baseline. No fitted parameter is renamed as a prediction: the SVD subspaces and losses are training supervision, while the reported metrics are identity-level matches on held-out test sets, so the central claim does not reduce to its inputs by construction. The self-citations to SapiensID and WebBody4M are appropriate because SapiensID is the baseline architecture and dataset; they are not used as an external uniqueness theorem or ansatz to forbid alternatives. The unverified disjointness of WebBody4M from test identities is a data-contamination and validity risk, and the Table 6 baselines trained on target-domain splits contradict the 'zero-shot' framing, but these concern external integrity and baseline fairness rather than circular derivation. The paper's stated limitation that teacher MLLM errors propagate to the student is an acknowledged boundedness caveat, not a circular step.

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

The method depends on the quality of the MLLM pseudo-labels and on the assumption that the training and test sets are disjoint. The free parameters are standard hyperparameters, but D_F and D_D directly control the amount of semantic knowledge transferred. No new physical or conceptual entities are introduced; the modules and losses are computational constructs.

free parameters (5)
  • D_F (fixed trait subspace dimension) = 6
    Number of principal components retained from the fixed soft-biometric text embeddings in ITA (Section 3.3, Eq. 3). It matches the six attribute categories in the prompt but is a design choice.
  • D_D (dynamic trait subspace dimension) = 5
    Number of principal components retained from the dynamic soft-biometric embeddings in TND (Section 3.3, Eq. 4). Matches five dynamic categories.
  • lambda_P (ITA projection weight) = 0.5
    Weight on the MSE term in the ITA loss (Section 3.3 and Appendix A.1). Chosen by hand.
  • Occlusion confidence threshold = 0.3
    Threshold for keypoint confidence to determine occlusion mask in K-SAH (Section 3.4, Eq. 8).
  • Temporal window size T = 5
    Window size for K-SAH during training (Section 4 and Appendix A.1). Ablated in Appendix A.6; performance saturates at T=5.
assumptions (4)
  • domain assumption The MLLM (Qwen2.5-VL) provides accurate fixed and dynamic soft-biometric attributes for WebBody4M images.
    The entire semantic distillation relies on the quality of these pseudo-labels (Section 3.3). The paper does not evaluate label accuracy.
  • domain assumption The text encoder embeddings of the attributes capture the semantic structure of the attributes.
    The embeddings are used to compute principal subspaces and Gram matrices (Eqs. 3-5). If the embeddings are uninformative, the losses are meaningless.
  • domain assumption WebBody4M identities are disjoint from test benchmark identities.
    Zero-shot evaluation assumes no training/test identity overlap (Section 4). No overlap analysis is provided.
  • standard math SVD of empirical covariance matrices yields stable principal components.
    Used in ITA and TND to define subspaces. Standard linear algebra, but the stability on mini-batches is not analyzed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SapiensID 2.0: Aligning Human Recognition Foundation Models with Human Perception." pith.science (2026). https://pith.science/paper/VCRA5REW

@misc{pith2026260810497,
  author       = {Pith},
  title        = {Pith review of: SapiensID 2.0: Aligning Human Recognition Foundation Models with Human Perception},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VCRA5REW}},
  note         = {Machine review of arXiv:2608.10497}
}
read the original abstract

While foundation models have significantly advanced human recognition across diverse modalities, they predominantly rely on static, geometric feature extraction. This approach fundamentally diverges from human perception. Consequently, current models often suffer from "semantic blindness," overfitting to transient noise while failing to leverage invariant soft biometrics, and struggle to capture temporal motion signatures. To bridge this gap, we propose SapiensID 2.0, a human recognition framework enriched with both semantic and temporal awareness. To overcome the lack of soft-biometric annotations, we transfer zero-shot semantic knowledge from Multimodal Large Language Models (MLLMs) into a discriminative embedding space. We resolve the dimensional mismatch between these spaces using Invariant Trait Alignment (ITA) to distill core persistent traits, and Transient Noise Disentanglement (TND) to decouple artifacts like clothing. Furthermore, we design a Kinematic Semantic Attention Head (K-SAH) that extends spatial attention across temporal windows. By tracking semantic patches over time, K-SAH captures rich kinematic signatures without requiring large-scale video datasets. Extensive experiments demonstrate that SapiensID 2.0 achieves state-of-the-art performance across image- and video-based person re-identification and gait recognition, while maintaining robust face recognition capabilities.

Figures

Figures reproduced from arXiv: 2608.10497 by the authors.

Figure 1
Figure 1. SapiensID [29], based on face recognition and person reID, mistakenly assigns high similarity to pairs, since it does not consider fixed soft biometrics like race and heavily relies on dynamic soft biometrics like clothing. Conversely, our approach, SapiensID 2.0, utilizes information extracted by MLLM from images to align fixed soft biometrics with biometric features and disentangle noise to achieve more robust hum… view at source ↗
Figure 2
Figure 2. Overall architecture of SapiensID 2.0. RetinaPatch (RP) converts the image into patches based on the landmarks and key points. ViT extracts features from these patches. K-SAH aggregates the features around each joint in images or videos. Additionally, we supervise the training via alignment with fixed soft biometrics and the disentanglement of dynamic soft biometrics. where K is the total number of semantic queries … view at source ↗
Figure 3
Figure 3. Semantic alignment via distillation. ITA aligns the principal components of the Y with the principal components of fixed soft biometric embeddings in the MLLM space. TND enforces that the principal components of the Y are orthogonal to those of the confounding factors like clothing, thereby minimizing their influence. To bridge the dimensional gap between the visual metric space and the language space during trainin… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Illustration of K-SAH. It extends static semantic attention to model local temporal dynamics over a temporal window W. Semantic queries Qt for the anchor frame t are derived via grid sampling from static embeddings, preserving spatial priors. Cross-frame attention betw…
Figure 5
Figure 5. Figure 5: Left: Compared to SapiensID, SapiensID 2.0 reduces the similarity between faces of different genders to below the verification threshold. Right: In retrieval, SapiensID returns another subject wearing similar clothing at rank-1. SapiensID 2.0 reduces the similarity bet…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

71 extracted references · 67 canonical work pages

  1. [1]

    A survey on soft biometrics for human identifi- cation

    Abdelgader Abdelwhab and Serestina Viriri. “A survey on soft biometrics for human identifi- cation”. In:Machine Learning and Biometrics(2018)

    work page 2018

  2. [2]

    Soft biometrics—combining body weight and fat measurements with fingerprint biometrics

    Heikki Ailisto, Elena Vildjiounaite, Mikko Lindholm, Satu-Marja Mäkelä, and Johannes Peltola. “Soft biometrics—combining body weight and fat measurements with fingerprint biometrics”. In:Pattern Recognition Letters(2006)

    work page 2006

  3. [3]

    GEFF: Improving Any Clothes- Changing Person ReID Model Using Gallery Enrichment With Face Features

    Daniel Arkushin, Bar Cohen, Shmuel Peleg, and Ohad Fried. “GEFF: Improving Any Clothes- Changing Person ReID Model Using Gallery Enrichment With Face Features”. In:WACV. 2024

    work page 2024

  4. [4]

    ViViT: A Video Vision Transformer

    Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lu ˇci´c, and Cordelia Schmid. “ViViT: A Video Vision Transformer”. In:ICCV. 2021

    work page 2021

  5. [5]

    Qwen2. 5-VL Technical Report

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, et al. “Qwen2. 5-VL Technical Report”. In:arXiv e-prints (2025), arXiv–2502

    work page 2025

  6. [6]

    Learning from Synchroniza- tion: Self-Supervised Uncalibrated Multi-View Person Association in Challenging Scenes

    Keqi Chen, Vinkle Srivastav, Didier Mutter, and Nicolas Padoy. “Learning from Synchroniza- tion: Self-Supervised Uncalibrated Multi-View Person Association in Challenging Scenes”. In: CVPR. 2025

    work page 2025

  7. [7]

    Beyond appearance: a semantic controllable self-supervised learning framework for human-centric visual tasks

    Weihua Chen, Xianzhe Xu, Jian Jia, Hao Luo, Yaohua Wang, Fan Wang, Rong Jin, and Xiuyu Sun. “Beyond appearance: a semantic controllable self-supervised learning framework for human-centric visual tasks”. In:CVPR. 2023

    work page 2023

  8. [8]

    MobileVLM V2: Faster and Stronger Baseline for Vision Language Model

    Xiangxiang Chu, Limeng Qiao, Xinyu Zhang, Shuang Xu, Fei Wei, Yang Yang, Xiaofei Sun, Yiming Hu, Xinyang Lin, Bo Zhang, and Chunhua Shen. “MobileVLM V2: Faster and Stronger Baseline for Vision Language Model”. In:arXiv preprint arXiv:2402.03766(2024). 10

    arXiv 2024

Show all 71 references
  1. [9]

    DKC: Differentiated Knowledge Consolidation for Cloth-Hybrid Lifelong Person Re-identification

    Zhenyu Cui, Jiahuan Zhou, and Yuxin Peng. “DKC: Differentiated Knowledge Consolidation for Cloth-Hybrid Lifelong Person Re-identification”. In:CVPR. 2025

    2025

  2. [10]

    An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. “An Image is Worth 16x16 Words: Transformers for Image Recognition a...

    2021

  3. [11]

    GIF: Generative Inspiration for Face Recognition at Scale

    Saeed Ebrahimi, Sahar Rahimi, Ali Dabouei, Srinjoy Das, Jeremy M Dawson, and Nasser M Nasrabadi. “GIF: Generative Inspiration for Face Recognition at Scale”. In:Proceedings of the Computer Vision and Pattern Recognition Conference. 2025

    2025

  4. [12]

    Exploring Deep Models for Practical Gait Recognition

    Chao Fan, Saihui Hou, Yongzhen Huang, and Shiqi Yu. “Exploring Deep Models for Practical Gait Recognition”. In:arXiv preprint arXiv:2303.03301(2023)

    2023 arXiv

  5. [13]

    Open- Gait: Revisiting Gait Recognition Towards Better Practicality

    Chao Fan, Junhao Liang, Chuanfu Shen, Saihui Hou, Yongzhen Huang, and Shiqi Yu. “Open- Gait: Revisiting Gait Recognition Towards Better Practicality”. In:CVPR. 2023

    2023

  6. [14]

    Skeletongait: Gait recognition using skeleton maps

    Chao Fan, Jingzhe Ma, Dongyang Jin, Chuanfu Shen, and Shiqi Yu. “Skeletongait: Gait recognition using skeleton maps”. In:AAAI. 2024

    2024

  7. [15]

    Clothes- changing person re-identification with rgb modality only

    Xinqian Gu, Hong Chang, Bingpeng Ma, Shutao Bai, Shiguang Shan, and Xilin Chen. “Clothes- changing person re-identification with rgb modality only”. In:CVPR. 2022

    2022

  8. [16]

    Multi-modal Multi-platform Person Re-Identification: Benchmark and Method

    Ruiyang Ha, Songyi Jiang, Bin Li, Bikang Pan, Yihang Zhu, Junjie Zhang, Xiatian Zhu, Shaogang Gong, and Jingya Wang. “Multi-modal Multi-platform Person Re-Identification: Benchmark and Method”. In:ICCV. 2025

    2025

  9. [17]

    Cargait: Cross- attention based re-ranking for gait recognition

    Gavriel Habib, Noa Barzilay, Or Shimshi, Rami Ben-Ari, and Nir Darshan. “Cargait: Cross- attention based re-ranking for gait recognition”. In:ICCV. 2025

    2025

  10. [18]

    Soft biometrics: A survey: Benchmark analysis, open challenges and recommendations

    Bilal Hassan, Ebroul Izquierdo, and Tomas Piatrik. “Soft biometrics: A survey: Benchmark analysis, open challenges and recommendations”. In:Multimedia Tools and Applications83.5 (2024), pp. 15151–15194

    2024

  11. [19]

    Labeled faces in the wild: A database forstudying face recognition in unconstrained environments

    Gary B Huang, Marwan Mattar, Tamara Berg, and Eric Learned-Miller. “Labeled faces in the wild: A database forstudying face recognition in unconstrained environments”. In:Workshop on faces in’Real-Life’Images: detection, alignment, and recognition. 2008

    2008

  12. [20]

    V ocabulary-Guided Gait Recognition

    Panjian Huang, Saihui Hou, Chunshui Cao, Xu Liu, and Yongzhen Huang. “V ocabulary-Guided Gait Recognition”. In:NeurIPS. 2025

    2025

  13. [21]

    Learning A Unified Template for Gait Recognition

    Panjian Huang, Saihui Hou, Junzhou Huang, and Yongzhen Huang. “Learning A Unified Template for Gait Recognition”. In:ICCV. 2025

    2025

  14. [22]

    VILLS : Video-Image Learning to Learn Semantics for Person Re-Identification

    Siyuan Huang, Ram Prabhakar Kathirvel, Yuxiang Guo, Rama Chellappa, and Cheng Peng. “VILLS : Video-Image Learning to Learn Semantics for Person Re-Identification”. In:WACV. 2025

    2025

  15. [23]

    Celebrities-ReID: A benchmark for clothes variation in long-term person re-identification

    Yan Huang, Qiang Wu, Jingsong Xu, and Yi Zhong. “Celebrities-ReID: A benchmark for clothes variation in long-term person re-identification”. In:IJCNN. 2019

    2019

  16. [24]

    Integrating faces, fingerprints, and soft biometric traits for user recognition

    Anil K Jain, Karthik Nandakumar, Xiaoguang Lu, and Unsang Park. “Integrating faces, fingerprints, and soft biometric traits for user recognition”. In:International Workshop on Biometric Authentication. Springer. 2004

    2004

  17. [25]

    Facial marks: Soft biometric for face recognition

    Anil K Jain and Unsang Park. “Facial marks: Soft biometric for face recognition”. In:ICIP. IEEE. 2009

    2009

  18. [26]

    On Denoising Walking Videos for Gait Recognition

    Dongyang Jin, Chao Fan, Jingzhe Ma, Jingkai Zhou, Weihua Chen, and Shiqi Yu. “On Denoising Walking Videos for Gait Recognition”. In:CVPR. 2025

    2025

  19. [27]

    Bridging the Sky and Ground: Towards View-Invariant Feature Learning for Aerial-Ground Person Re-Identification

    Wajahat Khalid, Bin Liu, Xulin Li, Muhammad Waqas, and Muhammad Sher Afgan. “Bridging the Sky and Ground: Towards View-Invariant Feature Learning for Aerial-Ground Person Re-Identification”. In:ICCV. 2025

    2025

  20. [28]

    Adaface: Quality adaptive margin for face recognition

    Minchul Kim, Anil K Jain, and Xiaoming Liu. “Adaface: Quality adaptive margin for face recognition”. In:CVPR. 2022

    2022

  21. [29]

    Sapiensid: Foundation for human recognition

    Minchul Kim, Dingqiang Ye, Yiyang Su, Feng Liu, and Xiaoming Liu. “Sapiensid: Foundation for human recognition”. In:CVPR. 2025

    2025

  22. [30]

    VIGFace: Virtual Identity Generation for Privacy-Free Face Recognition Dataset

    Minsoo Kim, Min-Cheol Sagong, Gi Pyo Nam, Junghyun Cho, and Ig-Jae Kim. “VIGFace: Virtual Identity Generation for Privacy-Free Face Recognition Dataset”. In:Proceedings of the IEEE/CVF International Conference on Computer Vision. 2025. 11

    2025

  23. [31]

    IDFace: Face Template Protection for Efficient and Secure Identification

    Sunpill Kim, Seunghun Paik, Chanwoo Hwang, Dongsoo Kim, Junbum Shin, and Jae Hong Seo. “IDFace: Face Template Protection for Efficient and Secure Identification”. In:CVPR. 2025

    2025

  24. [32]

    One-Shot Knowledge Transfer for Scalable Person Re-Identification

    Longhua Li, Lei Qi, and Xin Geng. “One-Shot Knowledge Transfer for Scalable Person Re-Identification”. In:ICCV. 2025

    2025

  25. [33]

    An In-Depth Exploration of Person Re-Identification and Gait Recognition in Cloth-Changing Conditions

    Weijia Li, Saihui Hou, Chunjie Zhang, Chunshui Cao, Xu Liu, Yongzhen Huang, and Yao Zhao. “An In-Depth Exploration of Person Re-Identification and Gait Recognition in Cloth-Changing Conditions”. In:CVPR. 2023

    2023

  26. [34]

    DIFFER: Disentangling Identity Features via Semantic Cues for Clothes-Changing Person Re-ID

    Xin Liang and Yogesh S Rawat. “DIFFER: Disentangling Identity Features via Semantic Cues for Clothes-Changing Person Re-ID”. In:CVPR. 2025

    2025

  27. [35]

    Farsight: A physics- driven whole-body biometric system at large distance and altitude

    Feng Liu, Ryan Ashbaugh, Nicholas Chimitt, Najmul Hassan, Ali Hassani, Ajay Jaiswal, Minchul Kim, Zhiyuan Mao, Christopher Perry, Zhiyuan Ren, et al. “Farsight: A physics- driven whole-body biometric system at large distance and altitude”. In:WACV. 2024

    2024

  28. [36]

    Person Recognition at Altitude and Range: Fusion of Face, Body Shape and Gait

    Feng Liu, Nicholas Chimitt, Lanqing Guo, Jitesh Jain, Aditya Kane, Minchul Kim, Wes Robbins, Yiyang Su, Dingqiang Ye, Xingguang Zhang, et al. “Person Recognition at Altitude and Range: Fusion of Face, Body Shape and Gait”. In:arXiv preprint arXiv:2505.04616 (2025)

    2025 arXiv

  29. [37]

    Learning Clothing and Pose Invariant 3D Shape Representation for Long-Term Person Re-Identification

    Feng Liu, Minchul Kim, ZiAng Gu, Anil Jain, and Xiaoming Liu. “Learning Clothing and Pose Invariant 3D Shape Representation for Long-Term Person Re-Identification”. In:ICCV. 2023

    2023

  30. [38]

    Distilling CLIP with Dual Guidance for Learning Discriminative Human Body Shape Representation

    Feng Liu, Minchul Kim, Zhiyuan Ren, and Xiaoming Liu. “Distilling CLIP with Dual Guidance for Learning Discriminative Human Body Shape Representation”. In:CVPR. 2024

    2024

  31. [39]

    Biometrics recognition using deep learning: A survey

    Shervin Minaee, Amirali Abdolrashidi, Hang Su, Mohammed Bennamoun, and David Zhang. “Biometrics recognition using deep learning: A survey”. In:Artificial Intelligence Review (2023)

    2023

  32. [40]

    Agedb: the first manually collected, in-the-wild age database

    Stylianos Moschoglou, Athanasios Papaioannou, Christos Sagonas, Jiankang Deng, Irene Kotsia, and Stefanos Zafeiriou. “Agedb: the first manually collected, in-the-wild age database”. In:CVPRW. 2017

    2017

  33. [41]

    Identity-Clothing Similarity Modeling for Unsupervised Clothing Change Person Re-Identification

    Zhiqi Pang, Junjie Wang, Lingling Zhao, and Chunyu Wang. “Identity-Clothing Similarity Modeling for Unsupervised Clothing Change Person Re-Identification”. In:CVPR. 2025

    2025

  34. [42]

    Face matching and retrieval using soft biometrics

    Unsang Park and Anil K Jain. “Face matching and retrieval using soft biometrics”. In:IEEE transactions on information forensics and security(2010)

    2010

  35. [43]

    Stylized-Face: A Million-level Stylized Face Dataset for Face Recognition

    Zhengyuan Peng, Jianqing Xu, Yuge Huang, Jinkun Hao, Shouhong Ding, Zhizhong Zhang, Xin Tan, and Lizhuang Ma. “Stylized-Face: A Million-level Stylized Face Dataset for Face Recognition”. In:ICCV. 2025

    2025

  36. [44]

    Frontal to profile face verification in the wild

    Soumyadip Sengupta, Jun-Cheng Chen, Carlos Castillo, Vishal M Patel, Rama Chellappa, and David W Jacobs. “Frontal to profile face verification in the wild”. In:WACV. 2016

    2016

  37. [45]

    LidarGait++: Learning Local Features and Size Awareness from LiDAR Point Clouds for 3D Gait Recognition

    Chuanfu Shen, Rui Wang, Lixin Duan, and Shiqi Yu. “LidarGait++: Learning Local Features and Size Awareness from LiDAR Point Clouds for 3D Gait Recognition”. In:CVPR. 2025

    2025

  38. [46]

    Large-Scale Spatio-Temporal Person Re-identification: Algorithms and Benchmark

    Xiujun Shu, Xiao Wang, Xianghao Zang, Shiliang Zhang, Yuanqi Chen, Ge Li, and Qi Tian. “Large-Scale Spatio-Temporal Person Re-identification: Algorithms and Benchmark”. In: TCSVT(2021)

    2021

  39. [47]

    DLCR: A Generative Data Expansion Framework via Diffusion for Clothes- Changing Person Re-ID

    Nyle Siddiqui, Florinel Alin Croitoru, Gaurav Kumar Nayak, Radu Tudor Ionescu, and Mubarak Shah. “DLCR: A Generative Data Expansion Framework via Diffusion for Clothes- Changing Person Re-ID”. In:WACV. 2025

    2025

  40. [48]

    Towards Robust- ness of Person Search against Corruptions

    Woojung Son, Yoonki Cho, Guoyuan An, Chanmi Lee, and Sung-Eui Yoon. “Towards Robust- ness of Person Search against Corruptions”. In:ICCV. 2025

    2025

  41. [49]

    Hamobe: Hierarchical and adaptive mixture of biometric experts for video-based person reid

    Yiyang Su, Yunping Shi, Feng Liu, and Xiaoming Liu. “Hamobe: Hierarchical and adaptive mixture of biometric experts for video-based person reid”. In:ICCV. 2025

    2025

  42. [50]

    Facial soft biometric features for forensic face recognition

    Pedro Tome, Ruben Vera-Rodriguez, Julian Fierrez, and Javier Ortega-Garcia. “Facial soft biometric features for forensic face recognition”. In:Forensic science international(2015)

    2015

  43. [51]

    FastVLM: Efficient Vision Encoding for Vision Language Models

    Pavan Kumar Anasosalu Vasu, Fartash Faghri, Chun-Liang Li, Cem Koc, Nate True, Albert Antony, Gokul Santhanam, James Gabriel, Peter Grasch, Oncel Tuzel, and Hadi Pouransari. “FastVLM: Efficient Vision Encoding for Vision Language Models”. In:CVPR. 2025. 12

    2025

  44. [52]

    SeCap: Self-Calibrating and Adaptive Prompts for Cross-view Person Re-Identification in Aerial-Ground Networks

    Shining Wang, Yunlong Wang, Ruiqi Wu, Bingliang Jiao, Wenxuan Wang, and Peng Wang. “SeCap: Self-Calibrating and Adaptive Prompts for Cross-view Person Re-Identification in Aerial-Ground Networks”. In:CVPR. 2025

    2025

  45. [53]

    Gait-X: Exploring X modality for Generalized Gait Recognition

    Zengbin Wang, Saihui Hou, Junjie Li, Xu Liu, Chunshui Cao, Yongzhen Huang, Siye Wang, and Man Zhang. “Gait-X: Exploring X modality for Generalized Gait Recognition”. In:ICCV. 2025

    2025

  46. [54]

    Person transfer GAN to bridge domain gap for person re-identification

    Longhui Wei, Shiliang Zhang, Wen Gao, and Qi Tian. “Person transfer GAN to bridge domain gap for person re-identification”. In:CVPR. 2018

    2018

  47. [55]

    Gait recognition from a single image using a phase-aware gait cycle reconstruction network

    Chi Xu, Yasushi Makihara, Xiang Li, Yasushi Yagi, and Jianfeng Lu. “Gait recognition from a single image using a phase-aware gait cycle reconstruction network”. In:ECCV. 2020

    2020

  48. [56]

    Self-Reinforcing Pro- totype Evolution with Dual-Knowledge Cooperation for Semi-Supervised Lifelong Person Re-Identification

    Kunlun Xu, Fan Zhuo, Jiangmeng Li, Xu Zou, and Jiahuan Zhou. “Self-Reinforcing Pro- totype Evolution with Dual-Knowledge Cooperation for Semi-Supervised Lifelong Person Re-Identification”. In:ICCV. 2025

    2025

  49. [57]

    Person re-identification by contour sketch under moderate clothing change

    Qize Yang, Ancong Wu, and Wei-Shi Zheng. “Person re-identification by contour sketch under moderate clothing change”. In:TPAMI(2019)

    2019

  50. [58]

    Bridging gait recognition and large language models sequence modeling

    Shaopeng Yang, Jilong Wang, Saihui Hou, Xu Liu, Chunshui Cao, Liang Wang, and Yongzhen Huang. “Bridging gait recognition and large language models sequence modeling”. In:CVPR. 2025

    2025

  51. [59]

    Good is Bad: Causality Inspired Cloth-debiasing for Cloth-changing Person Re-identification

    Zhengwei Yang, Meng Lin, Xian Zhong, Yu Wu, and Zheng Wang. “Good is Bad: Causality Inspired Cloth-debiasing for Cloth-changing Person Re-identification”. In:CVPR. 2023

    2023

  52. [60]

    Biggergait: Unlocking gait recognition with layer-wise representations from large vision models

    Dingqiang Ye, Chao Fan, Zhanbo Huang, Chengwen Luo, Jianqiang Li, Shiqi Yu, and Xiaom- ing Liu. “Biggergait: Unlocking gait recognition with layer-wise representations from large vision models”. In:NeurIPS. 2025

    2025

  53. [61]

    BigGait: Learning Gait Representation You Want by Large Vision Models

    Dingqiang Ye, Chao Fan, Jingzhe Ma, Xiaoming Liu, and Shiqi Yu. “BigGait: Learning Gait Representation You Want by Large Vision Models”. In:CVPR. 2024

    2024

  54. [62]

    LVFace: Progressive Cluster Optimization for Large Vision Models in Face Recognition

    Jinghan You, Shanglin Li, Yuanrui Sun, Jiangchuan Wei, Mingyu Guo, Chao Feng, and Jiao Ran. “LVFace: Progressive Cluster Optimization for Large Vision Models in Face Recognition”. In:ICCV. 2025

    2025

  55. [63]

    From poses to identity: Training-free person re-identification via feature centralization

    Chao Yuan, Guiwei Zhang, Changxiao Ma, Tianyi Zhang, and Guanglin Niu. “From poses to identity: Training-free person re-identification via feature centralization”. In:CVPR. 2025

    2025

  56. [64]

    Hap: Structure-aware masked image modeling for human-centric perception

    Junkun Yuan, Xinyu Zhang, Hao Zhou, Jian Wang, Zhongwei Qiu, Zhiyin Shao, Shaofeng Zhang, Sifan Long, Kun Kuang, Kun Yao, et al. “Hap: Structure-aware masked image modeling for human-centric perception”. In:NeurIPS. 2023

    2023

  57. [65]

    VIPerson: Flexibly Generating Virtual Identity for Person Re-Identification

    Xiao-Wen Zhang, Delong Zhang, Yi-Xing Peng, Zhi Ouyang, Jingke Meng, and Wei-Shi Zheng. “VIPerson: Flexibly Generating Virtual Identity for Person Re-Identification”. In: ICCV. 2025

    2025

  58. [66]

    Scalable person re-identification: A benchmark

    Liang Zheng, Liyue Shen, Lu Tian, Shengjin Wang, Jingdong Wang, and Qi Tian. “Scalable person re-identification: A benchmark”. In:ICCV. 2015

    2015

  59. [67]

    Cross-pose lfw: A database for studying cross-pose face recognition in unconstrained environments

    Tianyue Zheng and Weihong Deng. “Cross-pose lfw: A database for studying cross-pose face recognition in unconstrained environments”. In:Beijing University of Posts and Telecommuni- cations, Tech. Rep(2018)

    2018

  60. [68]

    Cross-age lfw: A database for studying cross- age face recognition in unconstrained environments

    Tianyue Zheng, Weihong Deng, and Jiani Hu. “Cross-age lfw: A database for studying cross- age face recognition in unconstrained environments”. In:arXiv preprint arXiv:1708.08197 (2017)

    2017 arXiv

  61. [69]

    A quality-guided mixture of score-fusion experts framework for human recognition

    Jie Zhu, Yiyang Su, Minchul Kim, Anil Jain, and Xiaoming Liu. “A quality-guided mixture of score-fusion experts framework for human recognition”. In:ICCV. 2025

    2025

  62. [70]

    Cross-covariate gait recognition: A benchmark

    Shinan Zou, Chao Fan, Jianbo Xiong, Chuanfu Shen, Shiqi Yu, and Jin Tang. “Cross-covariate gait recognition: A benchmark”. In:AAAI. 2024

    2024

  63. [71]

    ReID5o: Achieving Omni Multi-modal Person Re-identification in a Single Model

    Jialong Zuo, Yongtai Deng, Mengdan Tan, Rui Jin, Dongyue Wu, Nong Sang, Liang Pan, and Changxin Gao. “ReID5o: Achieving Omni Multi-modal Person Re-identification in a Single Model”. In:NeurIPS. 2025. 13 A Additional Experiments A.1 Additional Implementation Details Architectur...

    2025

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.