Pith. sign in

REVIEW 3 major objections 6 minor 67 references

Unity is Strength: Unifying Convolutional and Transformeral Features for Better Person Re-Identification

T0 review · 3 major / 6 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 A dual-branch framework that fuses CNN local features with Transformer global features via stacked cross-attention modules reports state-of-the-art person re-identification on three public benchmarks.

desk verdict A well-ablated CNN-Transformer fusion for ReID with plausible but unsecured SOTA gains; worth reviewing after multi-run and hyperparameter-selection fixes. read the letter →

arxiv 2412.17239 v1 pith:QMHLPVNZ submitted 2024-12-23 cs.CV cs.IR

Yuhao Wang , Pingping Zhang , Xuehu Liu , Zhengzheng Tu , Huchuan Lu This is my paper · ORCID
classification cs.CVcs.IR
keywords personre-identificationconvolutionalneuralnetworkvisiontransformerfeaturefusioncross-attentiondual-brancharchitecturebenchmarkevaluation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to show that person re-identification improves when the final deep features of a convolutional network and a vision transformer are explicitly fused, rather than when one backbone is modified to imitate the other. The proposed FusionReID extracts local CNN features and global Transformer features in parallel, aligns them, and repeatedly exchanges information through stacked dual-attention fusion modules. On Market1501, DukeMTMC, and MSMT17, the full model reports state-of-the-art or near-state-of-the-art mAP and Rank-1, with the largest gains on the hardest benchmark, MSMT17. If the results hold up, the work offers a general recipe: take any CNN and any Transformer, fuse their deep features with mutual cross-attention, and supervise all intermediate feature streams.

What carries the argument

The engine is the Heterogenous Transmission Module (HTM), a stackable block pairing a Shared Encoding Unit (SEU) with two Mutual Fusion Units (MFUs). SEU applies self-attention separately to each branch's tokens to sharpen local or global emphasis; each MFU runs cross-attention in which one branch's global token serves as query against the other branch's local patch tokens as keys and values. An earlier Local Refinement Unit (LRU) uses depthwise and pointwise convolutions to align channel counts and spatial sizes. Two stacked HTM layers, with six jointly supervised features, produce the final concatenated descriptor.

What would settle it

Run the MSMT17 experiment five times with different random seeds while choosing the fused dimension and HTM depth on a separate validation split; if the mean mAP does not exceed the best prior method or falls below 69.5%, the reported state-of-the-art claim would be falsified.

Watch

Extended reading notes

Core claim

FusionReID claims that late, symmetric fusion of CNN and Transformer features works better than early or one-directional combination. In the architecture, ResNet50 and ViT-B/16 each produce feature maps; these are aligned by depthwise and pointwise convolutions (LRU); then each Heterogenous Transmission Module runs self-attention (SEU) followed by two cross-attention fusions (MFU) in which the CNN global token reads Transformer local patches and the Transformer global token reads CNN local patches. Stacking two such modules and supervising six features with cross-entropy and triplet losses yields, on MSMT17, 69.5% mAP and 86.7% Rank-1, compared with 54.3%/78.1% for ResNet50 alone and 64.6%/83.2% for ViT-B/16* alone; on Market1501 it reports 91.7%/96.3% and on DukeMTMC 83.5%/91.0%. The paper interprets the consistent increments in ablations as evidence that each fusion component contributes and that CNN and Transformer features are genuinely complementary.

Load-bearing premise

The central claim rests on the assumption that the reported gains are reproducible and not artifacts of choosing the fused feature dimension and the number of fusion layers by inspecting test-set accuracy curves.

Editorial extensions

If this is right

  • The full framework sets a new reported state of the art on MSMT17: 69.5% mAP and 86.7% Rank-1 at 256x128 input, rising to 70.5% mAP and 87.3% Rank-1 at 384x128 with overlapping patches.
  • On Market1501 and DukeMTMC, it reports 91.7% mAP and 96.3% Rank-1 and 83.5% mAP and 91.0% Rank-1, matching or exceeding the best prior hybrid methods.
  • Ablations show each module matters: adding LRU, SEU, and MFU in sequence raises MSMT17 mAP from 67.5% with plain dual-branch extraction to 69.5% with the full module.
  • The framework is backbone-agnostic: pairing ResNet152 with ViT-B/16* reaches 71.8% mAP on MSMT17, and pairing ResNet50 with DeiT-B/16* reaches 69.7% mAP.
  • Because supervision is placed on six intermediate features rather than only the final descriptor, the design is compatible with standard ReID losses and test-time concatenation.

Reading between the lines

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

  • Editorial inference: A natural next test is whether the same late mutual-fusion block transfers to other retrieval domains, such as vehicle re-identification or product search, where local detail and global context also matter; the paper does not report such experiments.
  • Editorial inference: Because the architecture treats the two backbones as black boxes, replacing ImageNet-pretrained ViT with a self-supervised or CLIP-pretrained Transformer would reveal whether the gains come from architecture complementarity or from the specific pretraining.
  • Editorial inference: Because the paper selects the fused dimension and the number of stacked HTM layers from accuracy curves on the test set, an independent validation-split study would clarify how much of the reported margin survives honest hyperparameter selection.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes FusionReID, a two-branch person Re-identification framework that extracts features with a CNN (ResNet50) and a Vision Transformer (ViT-B/16), aligns them with Local Refinement Units (LRU), and fuses them via stacked Heterogenous Transmission Modules (HTM) composed of Shared Encoding Units (SEU) and Mutual Fusion Units (MFU). The model is trained with six cross-entropy and triplet losses and evaluated on Market1501, DukeMTMC, and MSMT17. The authors report state-of-the-art or near-state-of-the-art results on all three benchmarks, together with ablations for each component, different backbone combinations, computational cost analyses, and attention visualizations. Source code is publicly released.

Significance. If the reported results are reproducible, the paper provides a clean and modular two-branch fusion baseline and demonstrates complementary gains from combining CNN and Transformer features. The systematic ablations, the backbone-generality study in Table VI, and the release of source code are genuine strengths. However, the central empirical claim is currently under-supported: key structural hyperparameters are selected using test-set accuracy curves, only single-run numbers are reported, and the comparison against single-backbone state-of-the-art methods is confounded by model capacity. The architecture combines well-known attention building blocks, so the contribution is best viewed as an engineering improvement for ReID rather than a conceptual advance.

major comments (3)
  1. [Sec. IV-C/IV-D, Table I, Figs. 5-6] The central SOTA claim is not statistically supported. The number of stacked HTM layers (L=2) and the fused dimension (D=768) are selected from accuracy curves computed on the MSMT17 test set (Figs. 5 and 6), and the same test set is then used for the headline comparison in Table I. The decisive margins are small (91.7 vs 91.1 mAP against NFormer on Market1501; 69.5 vs 67.4 mAP against TransReID on MSMT17). Without multi-seed variance estimates or a validation-based hyperparameter selection protocol, the reported gains may be within run-to-run variability or reflect test-set overfitting. Please report mean and standard deviation over at least three seeds, or re-select hyperparameters on a held-out validation split before evaluating on the test set.
  2. [Sec. IV-D, Fig. 6 vs Table I] The paper states that the fused dimension D=1152 gives the best MSMT17 result (70.0% mAP, 86.8% Rank-1), yet all main comparisons in Table I use D=768, which yields 69.5% mAP. If D=1152 is abandoned for complexity reasons, that decision should be stated before Table I, and the D=1152 configuration should appear in the main comparison table. As written, the headline configuration is not the best model identified by the authors' own test-set exploration.
  3. [Sec. IV-C vs Sec. IV-D, Tables I and II] The comparison against single-backbone Transformer methods is confounded by model capacity. Table II shows that Method1 (DFE without DMF, i.e., simple concatenation of the two backbone features) already reaches 67.5 mAP on MSMT17 with 109.1M parameters, while TransReID, a single-branch ViT-B/16 model, reaches 67.4 mAP. The full FusionReID reaches 69.5 mAP with 153.8M parameters. Thus a large part of the apparent gain over single-backbone methods comes from adding a second backbone rather than from the proposed DMF. The paper should include the simple two-branch concatenation baseline directly in Table I and should temper the claim that the results isolate the effectiveness of the fusion modules.
minor comments (6)
  1. [Throughout] The word 'Transformeral' appears in the title, abstract, and body; it should be 'Transformer' or 'Transformer-based'. There are also misspellings such as 'convoluational' in Sec. III-B and 'intergrate' in Sec. IV-C.
  2. [Sec. III-C, Eq. (10)] Equation (10) is not a complete cross-entropy loss: it is missing the summation over classes and the leading minus sign. It should read L_ce = -Σ_j y_j log( exp(W_j f_c) / Σ_k exp(W_k f_c) ).
  3. [Sec. III-B, Eqs. (6)-(7)] The Query in each Mutual Fusion Unit is a single global token, so stating Q(c,0) ∈ R^{d×HtWt} is dimensionally inconsistent; the query should be a d×1 vector (or the text should explain how the global token is expanded across spatial positions).
  4. [Sec. IV-A] DukeMTMC is cited as [5], but reference [5] is the ViT paper; the dataset citation to Ristani et al. is missing.
  5. [Table VI] Table VI is difficult to parse because the rows for f'_c,L, f'_t,L, and f_a are merged with parameter and FLOP counts in the same cells. A conventional layout with separate rows and a separate Params/FLOPs column would improve readability.
  6. [Sec. IV-D, Fig. 6 discussion] The sentence about the 384-dimension configuration is ambiguous: the text says this configuration has the smallest number of parameters and FLOPs but then refers to a 'slight increase in model memory and computational complexity.' Please clarify what is being compared.

Circularity Check

1 steps flagged · score 6.0 of 10

The headline MSMT17 superiority claim is partially circular: the stacked-layer count and fused dimension are chosen from MSMT17 accuracy curves and then the same benchmark is used as the evidence in Table I; self-citations are frequent but not load-bearing.

  1. fitted input called prediction [Sec. IV-D ('Influence of the Stacked HTM' and 'Influence of Fused Dimensions in DMF'), Figs. 5-6, and Table I.]
    "Influence of the Stacked HTM. We conduct experiments on MSMT17 to verify the influence of the stacked HTM. ... However, the performance is saturated when stacked layers are more than 2. Thus, we set the layers to 2 by default. Influence of Fused Dimensions in DMF. Fig. 6 shows the influence of using different fused dimensions. ... To avoid excessive model complexity, we set 768 as the default dimension."

    The hyperparameters L (number of stacked HTMs) and D (fused dimension) are selected using accuracy curves measured on the MSMT17 benchmark (Figs. 5 and 6), and the resulting configuration is then reported on the same MSMT17 benchmark in Table I (69.5% mAP, 86.7% Rank-1) as evidence of superiority. The benchmark used for model selection is the same benchmark used to substantiate the central prediction, so the reported MSMT17 number is a model-selection artifact rather than an independent predictive result. The claim on MSMT17 is therefore partially circular even though the architecture itself is not defined from the benchmark outcome.

full rationale

FusionReID is an empirical architecture paper; its equations (1)-(12) define the DFE, LRU, SEU, MFU, and HTM modules from standard components (depthwise convolution, self-attention, cross-attention, losses) and do not analytically reduce the reported accuracy to an input. No uniqueness theorem is invoked and no ansatz is imported through a self-citation. The many self-citations (e.g., Refs. [13], [18]-[25], [54]) are used for background and context, not to justify the central fusion design or benchmark numbers. The significant circularity concern is the evaluation loop: the number of stacked HTM layers and the fused dimension are chosen from MSMT17 accuracy curves (Figs. 5-6) and the same benchmark then supplies the headline comparison in Table I. That makes the MSMT17 superiority claim partially circular by construction. Market1501 and DukeMTMC results are not tuned on those datasets, so the central claim retains independent content on those two benchmarks; this prevents a higher score. The absence of multi-seed variance estimates is a separate robustness limitation, not a circularity.

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

The central claim is an empirical performance result. It rests on domain assumptions about the complementary strengths of CNNs and Transformers and on the reliability of standard benchmarks. The only hand-chosen parameters are architectural hyperparameters (fused dimension, number of layers, weight-sharing scheme), and these are selected empirically. No new physical entities are introduced.

free parameters (4)
  • Fused dimension D = 768
    Chosen based on the mAP curve in Fig. 6; the paper notes 1152 gives the best accuracy but 768 is selected to limit complexity.
  • Number of stacked HTM layers = 2
    Chosen from Fig. 5; performance saturates beyond 2 layers.
  • Loss weights in Ltotal = 1 (equal for all six features)
    The total loss in Eq. (12) sums six cross-entropy and triplet losses without learned or tuned weighting.
  • Weight sharing policy = SEU shared, MFU unshared
    Selected via ablation in Tab. IV; the paper does not provide a theoretical justification.
assumptions (4)
  • domain assumption CNNs specialize in local features and Transformers in global features; fusing them yields better ReID representations.
    This premise motivates the whole fusion framework and is stated in Sec. I and repeated in Sec. IV-E.
  • domain assumption ImageNet-pretrained ResNet50 and ViT-B/16 backbones transfer to person ReID.
    The method relies on pretrained weights; no training from scratch is considered (Sec. IV-B).
  • domain assumption The benchmark datasets (Market1501, DukeMTMC, MSMT17) and metrics (mAP, Rank-1) are reliable proxies for ReID performance.
    The central claim is entirely empirical, resting on these public benchmarks.
  • standard math Standard attention layers, residual connections, LayerNorm, GeMP pooling, and the triplet/cross-entropy losses work as described in the cited literature.
    The method uses these off-the-shelf components without re-deriving them (Sec. III-B).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unity is Strength: Unifying Convolutional and Transformeral Features for Better Person Re-Identification." pith.science (2026). https://pith.science/paper/QMHLPVNZ

@misc{pith2026241217239,
  author       = {Pith},
  title        = {Pith review of: Unity is Strength: Unifying Convolutional and Transformeral Features for Better Person Re-Identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QMHLPVNZ}},
  note         = {Machine review of arXiv:2412.17239}
}
read the original abstract

Person Re-identification (ReID) aims to retrieve the specific person across non-overlapping cameras, which greatly helps intelligent transportation systems. As we all know, Convolutional Neural Networks (CNNs) and Transformers have the unique strengths to extract local and global features, respectively. Considering this fact, we focus on the mutual fusion between them to learn more comprehensive representations for persons. In particular, we utilize the complementary integration of deep features from different model structures. We propose a novel fusion framework called FusionReID to unify the strengths of CNNs and Transformers for image-based person ReID. More specifically, we first deploy a Dual-branch Feature Extraction (DFE) to extract features through CNNs and Transformers from a single image. Moreover, we design a novel Dual-attention Mutual Fusion (DMF) to achieve sufficient feature fusions. The DMF comprises Local Refinement Units (LRU) and Heterogenous Transmission Modules (HTM). LRU utilizes depth-separable convolutions to align deep features in channel dimensions and spatial sizes. HTM consists of a Shared Encoding Unit (SEU) and two Mutual Fusion Units (MFU). Through the continuous stacking of HTM, deep features after LRU are repeatedly utilized to generate more discriminative features. Extensive experiments on three public ReID benchmarks demonstrate that our method can attain superior performances than most state-of-the-arts. The source code is available at https://github.com/924973292/FusionReID.

Figures

Figures reproduced from arXiv: 2412.17239 by the authors.

Figure 1
Figure 1. Different unities of CNNs and Transformers. (a) Transformer is [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overall architecture of the proposed FusionReID. With CNN and Transformer backbones, the Dual-branch Feature Extraction (DFE) is used to extract [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Illustrations of our key modules. (a) Details of continuously stacked HTM; (b) SEU for deep feature enhancement; (c) MFU for deep feature fusion. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Various structures in HTM. (a) SEUs in front and MFUs in back; [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Effects of stacked HTM on MSMT17 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Effects of feature dimension D on MSMT17. However, the performance is saturated when stacked layers are more than 2. Thus, we set the layers to 2 by default. Influence of Fused Dimensions in DMF [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Visualization of different deep features with Grad-CAM. (a) Original images; Convolutional features (b) from ResNet50 only, (c) before LRU, (d) [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Visualization of attention weights in different heads. The first layer of (a) SEU and (b) MFU in the Transformer branch; The second layer of (c) [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: Attention weights in SEU and MFU. Top and bottom weights are [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

67 extracted references · 50 canonical work pages

  1. [1]

    Deep-person: Learning discriminative deep features for person re-identification,

    X. Bai, M. Yang, T. Huang, Z. Dou, R. Yu, and Y . Xu, “Deep-person: Learning discriminative deep features for person re-identification,” PR, vol. 98, p. 107036, 2020

    work page 2020

  2. [2]

    Learning to know where to see: A visibility-aware approach for occluded person re-identification,

    J. Yang, J. Zhang, F. Yu, X. Jiang, M. Zhang, X. Sun, Y .-C. Chen, and W.-S. Zheng, “Learning to know where to see: A visibility-aware approach for occluded person re-identification,” in ICCV, 2021, pp. 11 885–11 894

    work page 2021

  3. [3]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” NeurIPS, vol. 30, 2017

    2017

  4. [4]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in CVPR, 2016, pp. 770–778

    2016

  5. [5]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929 , 2020

    arXiv 2010

  6. [6]

    Visible thermal per- son re-identification via multi-branch modality residual complementary learning,

    L. Chen, R. Sun, Y . Yu, Y . Du, and X. Zhang, “Visible thermal per- son re-identification via multi-branch modality residual complementary learning,” IVC, p. 105201, 2024

    work page 2024

  7. [7]

    Robust Pseudo-label Learning with Neighbor Relation for Unsupervised Visible-Infrared Person Re-Identification

    X. Yin, J. Shi, Y . Zhang, Y . Lu, Z. Zhang, Y . Xie, and Y . Qu, “Robust pseudo-label learning with neighbor relation for unsupervised visible- infrared person re-identification,” arXiv preprint arXiv:2405.05613 , 2024

    work page Pith review arXiv 2024

  8. [8]

    Dual pseudo-labels interactive self-training for semi-supervised visible- infrared person re-identification,

    J. Shi, Y . Zhang, X. Yin, Y . Xie, Z. Zhang, J. Fan, Z. Shi, and Y . Qu, “Dual pseudo-labels interactive self-training for semi-supervised visible- infrared person re-identification,” in ICCV, 2023, pp. 11 218–11 228

    work page 2023

Show all 67 references
  1. [9]

    Multi-memory matching for unsupervised visible-infrared person re- identification,

    J. Shi, X. Yin, Y . Chen, Y . Zhang, Z. Zhang, Y . Xie, and Y . Qu, “Multi-memory matching for unsupervised visible-infrared person re- identification,” arXiv preprint arXiv:2401.06825 , 2024

    2024 arXiv

  2. [10]

    Abd-net: Attentive but diverse person re-identification,

    T. Chen, S. Ding, J. Xie, Y . Yuan, W. Chen, Y . Yang, Z. Ren, and Z. Wang, “Abd-net: Attentive but diverse person re-identification,” in ICCV, 2019, pp. 8351–8361

    2019

  3. [11]

    Batch dropblock network for person re-identification and beyond,

    Z. Dai, M. Chen, X. Gu, S. Zhu, and P. Tan, “Batch dropblock network for person re-identification and beyond,” in ICCV, 2019, pp. 3691–3701

    2019

  4. [12]

    High-order information matters: Learning relation and topology for occluded person re-identification,

    G. Wang, S. Yang, H. Liu, Z. Wang, Y . Yang, S. Wang, G. Yu, E. Zhou, and J. Sun, “High-order information matters: Learning relation and topology for occluded person re-identification,” in CVPR, 2020, pp. 6449–6458

    2020

  5. [13]

    Hat: Hierarchical aggregation transformers for person re-identification,

    G. Zhang, P. Zhang, J. Qi, and H. Lu, “Hat: Hierarchical aggregation transformers for person re-identification,” in ACM MM, 2021, pp. 516– 525

    2021

  6. [14]

    Relation-aware global attention for person re-identification,

    Z. Zhang, C. Lan, W. Zeng, X. Jin, and Z. Chen, “Relation-aware global attention for person re-identification,” in CVPR, 2020, pp. 3186–3195

    2020

  7. [15]

    Nformer: Robust person re-identification with neighbor transformer,

    H. Wang, J. Shen, Y . Liu, Y . Gao, and E. Gavves, “Nformer: Robust person re-identification with neighbor transformer,” in CVPR, 2022, pp. 7297–7307

    2022

  8. [16]

    Transreid: Transformer-based object re-identification,

    S. He, H. Luo, P. Wang, F. Wang, H. Li, and W. Jiang, “Transreid: Transformer-based object re-identification,” in ICCV, 2021, pp. 15 013– 15 022

    2021

  9. [17]

    Aaformer: Auto-aligned transformer for person re-identification,

    K. Zhu, H. Guo, S. Zhang, Y . Wang, G. Huang, H. Qiao, J. Liu, J. Wang, and M. Tang, “Aaformer: Auto-aligned transformer for person re-identification,” arXiv preprint arXiv:2104.00921 , 2021. IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 10

    2021 arXiv

  10. [18]

    Magic tokens: Select diverse tokens for multi-modal object re-identification,

    P. Zhang, Y . Wang, Y . Liu, Z. Tu, and H. Lu, “Magic tokens: Select diverse tokens for multi-modal object re-identification,” in CVPR, 2024, pp. 17 117–17 126

    2024

  11. [19]

    Top-reid: Multi- spectral object re-identification with token permutation,

    Y . Wang, X. Liu, P. Zhang, H. Lu, Z. Tu, and H. Lu, “Top-reid: Multi- spectral object re-identification with token permutation,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 38, no. 6, 2024, pp. 5758–5766

    2024

  12. [20]

    Other tokens matter: Exploring global and local features of vision transformers for object re-identification,

    Y . Wang, P. Zhang, D. Wang, and H. Lu, “Other tokens matter: Exploring global and local features of vision transformers for object re-identification,” CVIU, vol. 244, p. 104030, 2024

    2024

  13. [21]

    Tf-clip: Learning text-free clip for video-based person re-identification,

    C. Yu, X. Liu, Y . Wang, P. Zhang, and H. Lu, “Tf-clip: Learning text-free clip for video-based person re-identification,” in AAAI, vol. 38, no. 7, 2024, pp. 6764–6772

    2024

  14. [22]

    A video is worth three views: Trigeminal transformers for video-based person re- identification,

    X. Liu, P. Zhang, C. Yu, X. Qian, X. Yang, and H. Lu, “A video is worth three views: Trigeminal transformers for video-based person re- identification,” TITS, 2024

    2024

  15. [23]

    Deeply coupled convolution– transformer with spatial–temporal complementary learning for video- based person re-identification,

    X. Liu, C. Yu, P. Zhang, and H. Lu, “Deeply coupled convolution– transformer with spatial–temporal complementary learning for video- based person re-identification,” TNNLS, 2023

    2023

  16. [24]

    Video-based person re-identification with long short-term representation learning,

    X. Liu, P. Zhang, and H. Lu, “Video-based person re-identification with long short-term representation learning,” in ICIG. Springer, 2023, pp. 55–67

    2023

  17. [25]

    Learning progressive modality-shared transformers for effective visible-infrared person re-identification,

    H. Lu, X. Zou, and P. Zhang, “Learning progressive modality-shared transformers for effective visible-infrared person re-identification,” in AAAI, vol. 37, no. 2, 2023, pp. 1835–1843

    2023

  18. [26]

    Cvt: Introducing convolutions to vision transformers,

    H. Wu, B. Xiao, N. Codella, M. Liu, X. Dai, L. Yuan, and L. Zhang, “Cvt: Introducing convolutions to vision transformers,” in ICCV, 2021, pp. 22–31

    2021

  19. [27]

    Mobile-former: Bridging mobilenet and transformer,

    Y . Chen, X. Dai, D. Chen, M. Liu, X. Dong, L. Yuan, and Z. Liu, “Mobile-former: Bridging mobilenet and transformer,” in CVPR, 2022, pp. 5270–5279

    2022

  20. [28]

    Deep metric learning for person re-identification,

    D. Yi, Z. Lei, S. Liao, and S. Z. Li, “Deep metric learning for person re-identification,” in ICPR. IEEE, 2014, pp. 34–39

    2014

  21. [29]

    Beyond part models: Person retrieval with refined part pooling (and a strong convolutional baseline),

    Y . Sun, L. Zheng, Y . Yang, Q. Tian, and S. Wang, “Beyond part models: Person retrieval with refined part pooling (and a strong convolutional baseline),” in ECCV, 2018, pp. 480–496

    2018

  22. [30]

    Learning discriminative features with multiple granularities for person re-identification,

    G. Wang, Y . Yuan, X. Chen, J. Li, and X. Zhou, “Learning discriminative features with multiple granularities for person re-identification,” in ACM MM, 2018, pp. 274–282

    2018

  23. [31]

    Deep representation learning with part loss for person re-identification,

    H. Yao, S. Zhang, R. Hong, Y . Zhang, C. Xu, and Q. Tian, “Deep representation learning with part loss for person re-identification,” TIP, vol. 28, no. 6, pp. 2860–2871, 2019

    2019

  24. [32]

    A pose- sensitive embedding for person re-identification with expanded cross neighborhood re-ranking,

    M. S. Sarfraz, A. Schumann, A. Eberle, and R. Stiefelhagen, “A pose- sensitive embedding for person re-identification with expanded cross neighborhood re-ranking,” in CVPR, 2018, pp. 420–429

    2018

  25. [33]

    Pose-invariant embedding for deep person re-identification,

    L. Zheng, Y . Huang, H. Lu, and Y . Yang, “Pose-invariant embedding for deep person re-identification,” TIP, vol. 28, no. 9, pp. 4500–4509, 2019

    2019

  26. [34]

    Foreground- aware pyramid reconstruction for alignment-free occluded person re- identification,

    L. He, Y . Wang, W. Liu, H. Zhao, Z. Sun, and J. Feng, “Foreground- aware pyramid reconstruction for alignment-free occluded person re- identification,” in ICCV, 2019, pp. 8450–8459

    2019

  27. [35]

    Densely semantically aligned person re-identification,

    Z. Zhang, C. Lan, W. Zeng, and Z. Chen, “Densely semantically aligned person re-identification,” in CVPR, 2019, pp. 667–676

    2019

  28. [36]

    Bag of tricks and a strong baseline for deep person re-identification,

    H. Luo, Y . Gu, X. Liao, S. Lai, and W. Jiang, “Bag of tricks and a strong baseline for deep person re-identification,” in CVPRW, 2019, pp. 0–0

    2019

  29. [37]

    Learning deep feature representations with domain guided dropout for person re-identification,

    T. Xiao, H. Li, W. Ouyang, and X. Wang, “Learning deep feature representations with domain guided dropout for person re-identification,” in CVPR, 2016, pp. 1249–1258

    2016

  30. [38]

    Harmonious attention network for person re-identification,

    W. Li, X. Zhu, and S. Gong, “Harmonious attention network for person re-identification,” in CVPR, 2018, pp. 2285–2294

    2018

  31. [39]

    Diverse part dis- covery: Occluded person re-identification with part-aware transformer,

    Y . Li, J. He, T. Zhang, X. Liu, Y . Zhang, and F. Wu, “Diverse part dis- covery: Occluded person re-identification with part-aware transformer,” in CVPR, 2021, pp. 2898–2907

    2021

  32. [40]

    Adaptive high-frequency transformer for diverse wildlife re-identification,

    C. Li, S. Chen, and M. Ye, “Adaptive high-frequency transformer for diverse wildlife re-identification,” in ECCV. Springer, 2025, pp. 296– 313

    2025

  33. [41]

    Conformer: Convolution-augmented transformer for speech recognition,

    A. Gulati, J. Qin, C.-C. Chiu, N. Parmar, Y . Zhang, J. Yu, W. Han, S. Wang, Z. Zhang, Y . Wu et al., “Conformer: Convolution-augmented transformer for speech recognition,” arXiv preprint arXiv:2005.08100 , 2020

    2005 arXiv

  34. [42]

    Coatnet: Marrying convolution and attention for all data sizes,

    Z. Dai, H. Liu, Q. V . Le, and M. Tan, “Coatnet: Marrying convolution and attention for all data sizes,” NeurIPS, vol. 34, pp. 3965–3977, 2021

    2021

  35. [43]

    A semantic perception and cnn-transformer hybrid network for occluded person re-identification,

    Z. Gao, P. Chen, T. Zhuo, M. Liu, L. Zhu, M. Wang, and S. Chen, “A semantic perception and cnn-transformer hybrid network for occluded person re-identification,” TCSVT, vol. 34, no. 4, pp. 2010–2025, 2024

    2010

  36. [44]

    3d medical image segmentation using parallel transformers,

    Q. Yan, S. Liu, S. Xu, C. Dong, Z. Li, J. Q. Shi, Y . Zhang, and D. Dai, “3d medical image segmentation using parallel transformers,” PR, vol. 138, p. 109432, 2023

    2023

  37. [45]

    Ghostformer: Efficiently amalga- mated cnn-transformer architecture for object detection,

    X. Xie, D. Wu, M. Xie, and Z. Li, “Ghostformer: Efficiently amalga- mated cnn-transformer architecture for object detection,” PR, vol. 148, p. 110172, 2024

    2024

  38. [46]

    Rethinking the inception architecture for computer vision,

    C. Szegedy, V . Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, “Rethinking the inception architecture for computer vision,” in CVPR, 2016, pp. 2818–2826

    2016

  39. [47]

    In defense of the triplet loss for person re-identification,

    A. Hermans, L. Beyer, and B. Leibe, “In defense of the triplet loss for person re-identification,” arXiv preprint arXiv:1703.07737 , 2017

    2017 arXiv

  40. [48]

    Fine-tuning cnn image retrieval with no human annotation,

    F. Radenovi ´c, G. Tolias, and O. Chum, “Fine-tuning cnn image retrieval with no human annotation,” TPAMI, vol. 41, no. 7, pp. 1655–1668, 2018

    2018

  41. [49]

    Delving deep into rectifiers: Surpassing human-level performance on imagenet classification,

    K. He, X. Zhang, S. Ren, and J. Sun, “Delving deep into rectifiers: Surpassing human-level performance on imagenet classification,” in ICCV, 2015, pp. 1026–1034

    2015

  42. [50]

    Batch normalization: Accelerating deep network training by reducing internal covariate shift,

    S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” in ICML. pmlr, 2015, pp. 448–456

    2015

  43. [51]

    On layer normalization in the transformer architecture,

    R. Xiong, Y . Yang, D. He, K. Zheng, S. Zheng, C. Xing, H. Zhang, Y . Lan, L. Wang, and T. Liu, “On layer normalization in the transformer architecture,” in ICML. PMLR, 2020, pp. 10 524–10 533

    2020

  44. [52]

    Identity mappings in deep residual networks,

    K. He, X. Zhang, S. Ren, and J. Sun, “Identity mappings in deep residual networks,” in ECCV. Springer, 2016, pp. 630–645

    2016

  45. [53]

    Gaussian error linear units (gelus),

    D. Hendrycks and K. Gimpel, “Gaussian error linear units (gelus),” arXiv preprint arXiv:1606.08415, 2016

    2016 arXiv

  46. [54]

    Amulet: Aggregating multi-level convolutional features for salient object detection,

    P. Zhang, D. Wang, H. Lu, H. Wang, and X. Ruan, “Amulet: Aggregating multi-level convolutional features for salient object detection,” in ICCV, 2017, pp. 202–211

    2017

  47. [55]

    Scalable person re-identification: A benchmark,

    L. Zheng, L. Shen, L. Tian, S. Wang, J. Wang, and Q. Tian, “Scalable person re-identification: A benchmark,” in ICCV, 2015, pp. 1116–1124

    2015

  48. [56]

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

    L. Wei, S. Zhang, W. Gao, and Q. Tian, “Person transfer gan to bridge domain gap for person re-identification,” in CVPR, 2018, pp. 79–88

    2018

  49. [57]

    Imagenet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Imagenet: A large-scale hierarchical image database,” in CVPR. IEEE, 2009, pp. 248–255

    2009

  50. [58]

    Random erasing data augmentation,

    Z. Zhong, L. Zheng, G. Kang, S. Li, and Y . Yang, “Random erasing data augmentation,” in AAAI, vol. 34, no. 07, 2020, pp. 13 001–13 008

    2020

  51. [59]

    Omni-scale feature learning for person re-identification,

    K. Zhou, Y . Yang, A. Cavallaro, and T. Xiang, “Omni-scale feature learning for person re-identification,” in ICCV, 2019, pp. 3702–3712

    2019

  52. [60]

    Combined depth space based architec- ture search for person re-identification,

    H. Li, G. Wu, and W.-S. Zheng, “Combined depth space based architec- ture search for person re-identification,” in CVPR, 2021, pp. 6729–6738

    2021

  53. [61]

    Pose-guided feature dis- entangling for occluded person re-identification based on transformer,

    T. Wang, H. Liu, P. Song, T. Guo, and W. Shi, “Pose-guided feature dis- entangling for occluded person re-identification based on transformer,” in AAAI, vol. 36, no. 3, 2022, pp. 2540–2549

    2022

  54. [62]

    Seeing like a human: Asyn- chronous learning with dynamic progressive refinement for person re- identification,

    Q. Zhang, J. Lai, Z. Feng, and X. Xie, “Seeing like a human: Asyn- chronous learning with dynamic progressive refinement for person re- identification,” TIP, vol. 31, pp. 352–365, 2021

    2021

  55. [63]

    Incremental generative occlusion adversarial suppression network for person reid,

    C. Zhao, X. Lv, S. Dou, S. Zhang, J. Wu, and L. Wang, “Incremental generative occlusion adversarial suppression network for person reid,” TIP, vol. 30, pp. 4212–4224, 2021

    2021

  56. [64]

    Feature erasing and diffusion network for occluded person re-identification,

    Z. Wang, F. Zhu, S. Tang, R. Zhao, L. He, and J. Song, “Feature erasing and diffusion network for occluded person re-identification,” in CVPR, 2022, pp. 4754–4763

    2022

  57. [65]

    Tokens-to-token vit: Training vision transformers from scratch on imagenet,

    L. Yuan, Y . Chen, T. Wang, W. Yu, Y . Shi, Z.-H. Jiang, F. E. Tay, J. Feng, and S. Yan, “Tokens-to-token vit: Training vision transformers from scratch on imagenet,” in ICCV, 2021, pp. 558–567

    2021

  58. [66]

    Training data-efficient image transformers & distillation through attention,

    H. Touvron, M. Cord, M. Douze, F. Massa, A. Sablayrolles, and H. Jégou, “Training data-efficient image transformers & distillation through attention,” in ICML. PMLR, 2021, pp. 10 347–10 357

    2021

  59. [67]

    Grad-cam: Visual explanations from deep networks via gradient-based localization,

    R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-cam: Visual explanations from deep networks via gradient-based localization,” in ICCV, 2017, pp. 618–626

    2017

Pith tools

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