Pith. sign in

REVIEW 4 major objections 7 minor 70 references

Rethinking Vision Transformer for Large-Scale Fine-Grained Image Retrieval

T0 review · 4 major / 7 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read EET shows a pruned ViT-Small can cut fine-grained retrieval latency by 42.7% while improving 16-bit hash mAP on NABirds from 58.30 to 63.45.

desk verdict Solid efficiency story for fine-grained hashing, but Section IV-D's hash update equations are wrong as written and the CMBH comparison is buried. read the letter →

arxiv 2504.16691 v1 pith:EZGGK3Q5 submitted 2025-04-23 cs.MM

Xin Jiang , Hao Tang , Yonghua Pan , Zechao Li This is my paper · ORCID
classification cs.MM
keywords fine-grainedimageretrievalvisiontransformertokenpruninglearningtohashknowledgedistillationcontent-basedimportanceefficientinferencebinarycodes
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that the main barrier to deploying Vision Transformers in large-scale fine-grained image retrieval is not the architecture itself but the computation wasted on background and low-discriminative tokens. It proposes EET, a framework that progressively prunes those tokens and, during training only, transfers discriminative knowledge from a full ViT teacher into the pruned student, so inference stays fast. On six fine-grained benchmarks, EET reports ViT-Small inference latency cut by 42.7% (from 0.82 ms to 0.47 ms) while improving 16-bit hash-code mAP on NABirds from 58.30 to 63.45, surpassing the unpruned baseline and several dedicated hashing methods. If correct, the claim means efficiency and accuracy do not have to be traded off in fine-grained retrieval hashing.

What carries the argument

The engine of the method is the content-weighted token-importance score $M^l = \sum_{h=1}^{H} W^{h,l} \cdot A^{h,l}$, where $A^{h,l}$ is the class-token attention to each patch in layer $l$ and $W^{h,l}$ is the L2 norm of that head's output content normalized across heads. CTP keeps the top $N_\omega \cdot \mathrm{len}(M^l)$ tokens at layers 4, 8, and 10, progressively discarding background and low-discriminative patches. Two loss terms repair what pruning removes: DKT minimizes $1 - \cos(h_e, h_d)$ between the pruned student's hash codes and the teacher's, and DRG masks the most salient image region so the student is forced to learn secondary discriminative cues. The hash codes themselves come from a proxy-based two-step optimization borrowed from FISH, so the paper's contribution is the efficiency-and-distillation loop around a standard hash learner.

What would settle it

Take a fine-grained dataset with pixel-level annotations of discriminative regions (or use the paper's own failure images) and compute recall@kept: the fraction of annotated discriminative patches among the tokens EET retains at each pruning stage. If recall is low yet an oracle pruner that keeps the annotated patches scores higher mAP at equal latency, the CTP importance score is not the cause of the accuracy gain and the method's central mechanism is falsified.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that a ViT's redundant tokens can be productively removed rather than tolerated: ranking tokens by an attention score weighted by the L2 norm of each head's output content identifies discriminative patches, and a training-time distillation pair—cosine-distance hash-code transfer from the teacher plus a masked-image region-guidance loss—restores and even exceeds the accuracy lost to pruning. The result is a hashing model (EET) that, at 16-bit code length, raises NABirds mAP from 58.30 to 63.45 and cuts latency from 0.82 ms to 0.47 ms relative to ViT-Small, and that matches or beats the unpruned baseline across CUB-200-2011, Stanford Cars, VegFru, Food101, and iNat2017 at multiple code lengths.

Load-bearing premise

The load-bearing assumption is that ranking image patches by the L2 magnitude of each attention head's output content, folded into class-attention scores, reliably picks out the subtle discriminative regions of fine-grained objects; if that ranking prunes informative tokens, retrieval accuracy degrades, and the paper's own ablations and Figure 8 show exactly such failures on Stanford Cars and background-cluttered images.

Editorial extensions

If this is right

  • A pruned ViT-Small can replace the full ViT-Small in fine-grained hashing pipelines: latency drops from 0.82 ms to 0.47 ms with retrieval mAP equal or better on the reported datasets.
  • The speedup is not paid for in hash-code quality: EET's gains over the baseline appear at 16-, 32-, 48-, and 64-bit codes, and the largest relative gain is at 16 bits on NABirds (+5.15 mAP).
  • Pruning positions and ratios matter hierarchically: inserting CTP at layers 4, 8, and 10 with retention ratios 1/2, 1/2, and 1/4 gives the best accuracy-efficiency balance, while pruning only at layer 10 is nearly as accurate but slower.
  • Content-weighted class attention is a better token-ranking signal than raw class attention for fine-grained images; replacing it with the raw score lowers mAP on all three ablation datasets.
  • Because DKT and DRG are training-only, the discriminative transfer is cost-free at inference, so the efficiency gain of pruning is kept end-to-end.

Reading between the lines

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

  • The CTP ranking heuristic has not been shown to generalize to objects whose discriminative parts are subtle and low-response; the paper's own Figure 8 shows it failing on car fronts, headlights, and logos, and its ablations show CTP alone dropping Stanford Cars 16-bit mAP from 82.85 to 79.96. A natural next test is replacing the L2-content weight with a learned saliency predictor and measuring whe
  • DKT aligns continuous hash embeddings, not logits or intermediate features, so the teacher-student pair is architecture-asymmetric; this suggests the same transfer could run from a CNN or a larger ViT teacher into any efficient student, which the paper does not test.
  • DRG masks the globally most salient region rather than the per-class discriminative region; making the mask class-aware or adaptively sized could reduce the background-confusion failures the paper reports and is a direct, cheap ablation to try.
  • The paper frames progressive pruning as mimicking human global-to-local attention, but the mechanism is simpler: conditional computation plus distillation. If the distillation terms were removed, the model would be just a faster, weaker ViT, which is exactly what the ablation table shows.
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 / 7 minor

Summary. The paper proposes EET, an efficient ViT-based hashing framework for large-scale fine-grained image retrieval. EET consists of three components: content-based token pruning (CTP), which progressively removes low-discriminative tokens at fixed transformer layers using attention weighted by per-head output-content norms; discriminative knowledge transfer (DKT), which aligns the student hash codes with a heavier teacher ViT; and discriminative region guidance (DRG), which masks salient regions during training to force attention to subtler cues. The hash codes are learned through an alternating proxy-based optimization following FISH, and only the pruned student is used at inference. Experiments on CUB-200-2011, Stanford Cars, NABirds, VegFru, Food101, and iNat2017 report mAP, PR curves, GFLOPs, and latency, with the headline result that EET reduces ViT-Small latency by 42.7% and improves 16-bit NABirds mAP from 58.30 to 63.45.

Significance. If the reported results hold, EET is a practically useful combination of token pruning and distillation for fine-grained hashing: it is the first ViT-based method in this comparison to address both the quadratic attention cost and the fine-grained discriminative bottleneck, and the paper includes publicly available code, ablation studies, latency measurements, and experiments on six datasets. The central efficiency claim is supported by Table V, where CTP cuts latency from 0.82 ms to 0.47 ms. However, the significance is currently reduced by two technical problems in the hash-code optimization (Eqs. (14) and (16)), by the omission of the strongest competitor CMBH from the main comparison tables, and by the absence of variance reporting for hyperparameters tuned on a single dataset. With corrections and a more transparent comparison, this would be a solid contribution to the fine-grained retrieval and efficient-ViT literature.

major comments (4)
  1. [Section IV-D, Eq. (14)] The closed-form update for P is algebraically incorrect. Minimizing ||Y - PV||_F^2 with fixed V gives P = YV^T(VV^T)^{-1}, not P = YV^T(YV^T)^{-1}. As written, the formula is dimensionally invalid unless C = k and is wrong in general. Since the optimized P and the subsequent alternating updates produce the hash-code target B used in Eq. (17), this error makes the described method non-reproducible from the text; please correct the equation and verify the derivation.
  2. [Section IV-D, Eq. (16)] Minimizing ||B - RV||_F^2 over B is equivalent to maximizing Tr(B^T(RV)), not minimizing it. The sign in Eq. (16) is therefore flipped; B = sign(RV) is the argmax, not the argmin. This is not cosmetic: the B matrix is the training target for L_hash in Eq. (17), so the displayed optimization would generate inverted hash codes. Please fix the sign or clarify that Eq. (16) is intended as a maximization.
  3. [Tables I–IV and Section V-C3] CMBH is omitted from the main comparisons (Tables I–III) and appears only in Table IV, where it outperforms EET on nearly every entry (e.g., CUB 16-bit 84.07 vs. 75.02; VegFru 24-bit 88.63 vs. 86.94; Food101 32-bit 89.28 vs. 88.31). The statement in Section V-C3 that EET 'slightly lags behind CMBH' understates these gaps. The paper should either include CMBH in the headline tables or explicitly qualify the state-of-the-art claim.
  4. [Section V-B and Figures 4–5] The hyperparameters beta, sigma, and the pruning ratios are tuned on CUB-200-2011 and then applied to all other datasets, but the paper reports no standard deviations or repeated runs. Since the main cross-dataset claims (especially NABirds and the large-scale benchmarks) depend on these choices, please report variance across seeds and show at least one sensitivity analysis on a second dataset to support the claim that the settings transfer.
minor comments (7)
  1. [Section IV-C2] There is a typo in the text: 'As aforementioned, the the FGIR task' should read 'the FGIR task'.
  2. [Section V-C4 heading] The heading 'Ablatuion Studies' should be spelled 'Ablation Studies'.
  3. [Section IV-D] The sentence 'The learning process of the hash code is divided into two steps' is repeated verbatim twice in consecutive lines; please remove the duplicate.
  4. [Section IV-B and Table VIII] The notation for pruning ratios is inconsistent: Section IV-B uses N_omega, Section V-B uses gamma_j, and Table VIII uses (N1, N2, N3). Please unify the notation.
  5. [Section IV-C1, Eq. (9)] The identity hamm(b_i,b_j) = (k/2)(1 - cos(b_i,b_j)) is exact only for binary vectors b_i, b_j; the approximation with continuous hash codes h_i, h_j should be stated more carefully, since the cosine distance between continuous vectors does not exactly equal the Hamming distance after sign binarization.
  6. [Section IV-D, Eq. (13)] The rotation matrix R is declared as R in R^{r x r}, but it multiplies V in R^{k x n}; please define r = k or replace r with k throughout the optimization.
  7. [Section IV-F, Eq. (20)] The order of the heads in b_q = sign(FChash(FCcls(E_q_class))) appears reversed relative to the figure and the text, which describe the hash head as a projection of the final class token; please confirm whether FC_cls is applied before FC_hash or whether this is a typo.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the token-pruning heuristic and the two transfer losses are training objectives evaluated against external benchmarks, not quantities implied by their own definitions.

full rationale

The central claims are that CTP reduces latency by pruning low-importance tokens and that DKT/DRG restore discriminative power. CTP's importance score M^l = sum_h W^{h,l} · A^{h,l} (Eq. (8)) is a heuristic computed from L2 norms and class-attention scores; it is not derived from the retrieval mAP it is used to predict. DKT (Eq. (10)) is a cosine-distance distillation loss between teacher and student hash codes, and DRG (Eqs. (11)-(12)) is a masked-image classification loss; both are regularizers applied during training and evaluated on held-out test splits, so they cannot reduce to the reported test mAP by construction. The hash-code target B is obtained by a standard alternating optimization (Section IV-D) over the training labels and features, and the final hash function is learned from that target (Eq. (17)); this is conventional supervised hashing, not a fitted-parameter-calling-itself-prediction. The paper's self-citations (DVF, DAHN-ET) are used as comparison baselines and related work, not as load-bearing justification for EET's design. The limitations section concedes empirical failure cases of CTP, which further indicates the ablation comparisons are genuine rather than forced. No equation in the paper reduces to its own input by definition, and no uniqueness claim is imported from the authors' prior work. Separately, Eqs. (14) and (16) contain algebraic errors that undermine reproducibility, but algebraic incorrectness is not circularity.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The central claim rests on standard pretrained ViT components, a heuristic token-importance scoring rule, distillation assumptions, and several dataset-specific hyperparameters. There are no invented physical entities, but the method's generality depends on the pruning-ratio and loss-weight choices transferred from CUB to all other datasets.

free parameters (4)
  • pruning ratios (gamma_1, gamma_2, gamma_3 / Nomega) = 1/2, 1/2, 1/4
    Retain fraction of tokens after layers 4, 8, 10; selected from Table VIII as the best effectiveness-efficiency trade-off, then fixed on all datasets.
  • beta = 0.1
    Weight of classification and DRG losses in Eq. (19); tuned on CUB-200-2011 (Figure 4) where larger beta degrades mAP, then fixed for all datasets.
  • sigma = 1.0
    Weight of DKT loss in Eq. (19); chosen from robustness analysis on CUB (Figure 5).
  • K in DRG (number of masked salient locations) = not specified
    Eq. (11) uses TopK(.) but the paper never states K or how it scales across datasets; this is an undocumented tunable.
assumptions (6)
  • domain assumption Pre-trained ImageNet-1K ViT-Small provides a suitable initialization and teacher for fine-grained hashing.
    Used in all experiments (Section V-B); if the pretrained backbone were not transferable, the comparisons would change.
  • domain assumption Token importance can be ranked by combining class-token attention with normalized per-head output magnitudes (Eqs. 6-8).
    This is the founding heuristic of CTP; the paper provides no theoretical justification, only empirical ablations.
  • domain assumption Knowledge distillation from a full ViT to the pruned student preserves or restores fine-grained discriminative information.
    DKT and DRG in Section IV-C assume the teacher's hash codes and saliency maps are reliable guides; no independent verification beyond mAP tables.
  • domain assumption Proxy-based hash learning as described in FISH (Eqs. 13-17) is correct and transferable.
    The paper reproduces the FISH optimization, including Eq. (14); the reproduction contains apparent typos, so reliance on the external baseline is load-bearing.
  • ad hoc to paper The chosen pruning positions {4, 8, 10} are appropriate across all datasets.
    Justified only by Table VII ablations; no principle determines these layers.
  • standard math ViT forward pass equations (Eqs. 1-3) are standard building blocks with no re-derivation needed.
    The paper relies on standard ViT definitions; no proof is given, consistent with common practice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Rethinking Vision Transformer for Large-Scale Fine-Grained Image Retrieval." pith.science (2026). https://pith.science/paper/EZGGK3Q5

@misc{pith2026250416691,
  author       = {Pith},
  title        = {Pith review of: Rethinking Vision Transformer for Large-Scale Fine-Grained Image Retrieval},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EZGGK3Q5}},
  note         = {Machine review of arXiv:2504.16691}
}
read the original abstract

Large-scale fine-grained image retrieval (FGIR) aims to retrieve images belonging to the same subcategory as a given query by capturing subtle differences in a large-scale setting. Recently, Vision Transformers (ViT) have been employed in FGIR due to their powerful self-attention mechanism for modeling long-range dependencies. However, most Transformer-based methods focus primarily on leveraging self-attention to distinguish fine-grained details, while overlooking the high computational complexity and redundant dependencies inherent to these models, limiting their scalability and effectiveness in large-scale FGIR. In this paper, we propose an Efficient and Effective ViT-based framework, termed \textbf{EET}, which integrates token pruning module with a discriminative transfer strategy to address these limitations. Specifically, we introduce a content-based token pruning scheme to enhance the efficiency of the vanilla ViT, progressively removing background or low-discriminative tokens at different stages by exploiting feature responses and self-attention mechanism. To ensure the resulting efficient ViT retains strong discriminative power, we further present a discriminative transfer strategy comprising both \textit{discriminative knowledge transfer} and \textit{discriminative region guidance}. Using a distillation paradigm, these components transfer knowledge from a larger ``teacher'' ViT to a more efficient ``student'' model, guiding the latter to focus on subtle yet crucial regions in a cost-free manner. Extensive experiments on two widely-used fine-grained datasets and four large-scale fine-grained datasets demonstrate the effectiveness of our method. Specifically, EET reduces the inference latency of ViT-Small by 42.7\% and boosts the retrieval performance of 16-bit hash codes by 5.15\% on the challenging NABirds dataset.

Figures

Figures reproduced from arXiv: 2504.16691 by the authors.

Figure 1
Figure 1. (a) Coarse-grained images: Significant visual differences between [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed framework, which comprises three core components: [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Precision-Recall curves of EET and state-of-the-art methods on the three datasets. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The hyper-parameter analysis of β on CUB-200-2011 with code bits from 16 to 64. 0.2 0.4 0.6 0.8 1.0 72.0 74.0 76.0 78.0 80.0 82.0 84.0 86.0 mAP (%) 16bits 32bits 48bits 64bits [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: The hyper-parameter analysis of σ on CUB-200-2011 with code bits from 16 to 64. constant. The figure indicates that performance degrades as β increases, with the worst results at β = 1.0. This occurs because a high β value hinders the overall optimization of the model,…
Figure 6
Figure 6. Figure 6: Examples of top 10 retrieval samples of the proposed EET on the CUB-200-2011. The retrieval images with green boxes are the correct ones, and [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Visualization results of token pruning for samples from CUB-200- [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Failure cases of the CTP on Stanford Cars and CUB-200-2011 datasets. [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

70 extracted references · 60 canonical work pages

  1. [1]

    Fine-grained image analysis with deep learning: A sur- vey,

    X. Wei, Y . Song, O. M. Aodha, J. Wu, Y . Peng, J. Tang, J. Yang, and S. J. Belongie, “Fine-grained image analysis with deep learning: A sur- vey,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 12, pp. 8927–8948, 2022

    work page 2022

  2. [2]

    Deep collaborative embedding for social image understanding,

    Z. Li, J. Tang, and T. Mei, “Deep collaborative embedding for social image understanding,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 41, no. 9, pp. 2070–2083, 2019

    work page 2019

  3. [3]

    3d object representations for fine-grained categorization,

    J. Krause, M. Stark, J. Deng, and L. Fei-Fei, “3d object representations for fine-grained categorization,” in Proceedings of the IEEE Interna- tional Conference on Computer Vision Workshops , 2013, pp. 554–561

    work page 2013

  4. [4]

    The caltech-ucsd birds-200-2011 dataset,

    C. Wah, S. Branson, P. Welinder, P. Perona, and S. Belongie, “The caltech-ucsd birds-200-2011 dataset,” 2011

    2011

  5. [5]

    Dvf: Advancing robust and accurate fine-grained image retrieval with retrieval guidelines,

    X. Jiang, H. Tang, R. Yan, J. Tang, and Z. Li, “Dvf: Advancing robust and accurate fine-grained image retrieval with retrieval guidelines,” in Proceedings of the 32nd ACM International Conference on Multimedia , 2024, pp. 2379–2388

    work page 2024

  6. [6]

    Hypergraph-induced semantic tuplet loss for deep metric learning,

    J. Lim, S. Yun, S. Park, and J. Y . Choi, “Hypergraph-induced semantic tuplet loss for deep metric learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2022, pp. 212– 222

    work page 2022

  7. [7]

    No fuss distance metric learning using proxies,

    Y . Movshovitz-Attias, A. Toshev, T. K. Leung, S. Ioffe, and S. Singh, “No fuss distance metric learning using proxies,” in Proceedings of the IEEE International Conference on Computer Vision, 2017, pp. 360–368

    2017

  8. [8]

    One loss for all: Deep hashing with a single cosine similarity based learning objective,

    J. T. Hoe, K. W. Ng, T. Zhang, C. S. Chan, Y . Song, and T. Xiang, “One loss for all: Deep hashing with a single cosine similarity based learning objective,” in Proceedings of the Conference on Neural Information Processing Systems, 2021, pp. 24 286–24 298

    work page 2021

Show all 70 references
  1. [9]

    Global meets local: Dual activation hashing network for large-scale fine-grained image retrieval,

    X. Jiang, H. Tang, and Z. Li, “Global meets local: Dual activation hashing network for large-scale fine-grained image retrieval,” IEEE Transactions on Knowledge and Data Engineering , pp. 1–14, 2024

    2024

  2. [10]

    Deep polarized network for supervised learning of accurate binary hashing codes,

    L. Fan, K. W. Ng, C. Ju, T. Zhang, and C. S. Chan, “Deep polarized network for supervised learning of accurate binary hashing codes,” in Proceedings of the International Joint Conference on Artificial Intelli- gence, 2020, pp. 825–831

    2020

  3. [11]

    Central similarity quantization for efficient image and video retrieval,

    L. Yuan, T. Wang, X. Zhang, F. E. H. Tay, Z. Jie, W. Liu, and J. Feng, “Central similarity quantization for efficient image and video retrieval,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2020, pp. 3080–3089

    2020

  4. [12]

    SEMICON: A learning-to-hash solution for large-scale fine-grained image retrieval,

    Y . Shen, X. Sun, X. Wei, Q. Jiang, and J. Yang, “SEMICON: A learning-to-hash solution for large-scale fine-grained image retrieval,” in Proceedings of the European Conference on Computer Vision , 2022, pp. 531–548

    2022

  5. [13]

    Attribute-aware deep hashing with self-consistency for large-scale fine-grained image retrieval,

    X. Wei, Y . Shen, X. Sun, P. Wang, and Y . Peng, “Attribute-aware deep hashing with self-consistency for large-scale fine-grained image retrieval,” IEEE Transactions on Pattern Analysis and Machine Intelli- gence, vol. 45, no. 11, pp. 13 904–13 920, 2023

    2023

  6. [14]

    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, J. Uszkoreit, and N. Houlsby, “An image is worth 16x16 words: Transformers for image recognition at scale,” in Proceedings of the International Con...

    2021

  7. [15]

    Training data-efficient image transformers & distillation through attention,

    H. Touvron, M. Cord, M. Douze, F. Massa, A. Sablayrolles, and H. J ´egou, “Training data-efficient image transformers & distillation through attention,” in Proceedings of the International Conference on Machine Learning, 2021, pp. 10 347–10 357

    2021

  8. [16]

    Msvit: training multiscale vision transformers for image retrieval,

    X. Li, J. Yu, S. Jiang, H. Lu, and Z. Li, “Msvit: training multiscale vision transformers for image retrieval,” IEEE Transactions on Multimedia , 2023

    2023

  9. [17]

    Swinfghash: Fine-grained image retrieval via transformer-based hashing network

    D. Lu, J. Wang, Z. Zeng, B. Chen, S. Wu, and S.-T. Xia, “Swinfghash: Fine-grained image retrieval via transformer-based hashing network.” in Proceedings of the British Machine Vision Conference , 2021, pp. 432– 444

    2021

  10. [18]

    No matter how: Top-down effects of verbal and semantic category knowledge on early visual perception,

    M. Maier and R. Abdel Rahman, “No matter how: Top-down effects of verbal and semantic category knowledge on early visual perception,” Cognitive, Affective, & Behavioral Neuroscience , vol. 19, pp. 859–876, 2019

    2019

  11. [19]

    Learning attention-guided pyrami- dal features for few-shot fine-grained recognition,

    H. Tang, C. Yuan, Z. Li, and J. Tang, “Learning attention-guided pyrami- dal features for few-shot fine-grained recognition,” Pattern Recognition, vol. 130, p. 108792, 2022

    2022

  12. [20]

    Divide-and-conquer: Confluent triple-flow network for rgb-t salient object detection,

    H. Tang, Z. Li, D. Zhang, S. He, and J. Tang, “Divide-and-conquer: Confluent triple-flow network for rgb-t salient object detection,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 47, no. 3, pp. 1958–1974, 2025

    1958

  13. [21]

    Adaptive token sampling for efficient vision transformers,

    M. Fayyaz, S. A. Koohpayegani, F. R. Jafari, S. Sengupta, H. R. V . Joze, E. Sommerlade, H. Pirsiavash, and J. Gall, “Adaptive token sampling for efficient vision transformers,” in Proceedings of the European Conference on Computer Vision , 2022, pp. 396–414

    2022

  14. [22]

    Token merging: Your vit but faster,

    D. Bolya, C. Fu, X. Dai, P. Zhang, C. Feichtenhofer, and J. Hoffman, “Token merging: Your vit but faster,” inProceedings of the International Conference on Learning Representations , 2023

    2023

  15. [23]

    Dynamicvit: Efficient vision transformers with dynamic token sparsification,

    Y . Rao, W. Zhao, B. Liu, J. Lu, J. Zhou, and C. Hsieh, “Dynamicvit: Efficient vision transformers with dynamic token sparsification,” in Pro- ceedings of the Conference on Neural Information Processing Systems , 2021, pp. 13 937–13 949

    2021

  16. [24]

    Bilinear CNN models for fine- grained visual recognition,

    T. Lin, A. RoyChowdhury, and S. Maji, “Bilinear CNN models for fine- grained visual recognition,” in Proceedings of the IEEE International Conference on Computer Vision , 2015, pp. 1449–1457

    2015

  17. [25]

    Compact bilinear pooling,

    Y . Gao, O. Beijbom, N. Zhang, and T. Darrell, “Compact bilinear pooling,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 317–326

    2016

  18. [26]

    Hierarchical bilinear pooling for fine-grained visual recognition,

    C. Yu, X. Zhao, Q. Zheng, P. Zhang, and X. You, “Hierarchical bilinear pooling for fine-grained visual recognition,” in Proceedings of the European Conference on Computer Vision , 2018, pp. 595–610

    2018

  19. [27]

    Deep LAC: deep localization, alignment and classification for fine-grained recognition,

    D. Lin, X. Shen, C. Lu, and J. Jia, “Deep LAC: deep localization, alignment and classification for fine-grained recognition,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2015, pp. 1666–1674

    2015

  20. [28]

    Boosting few-shot fine-grained recognition with background suppression and foreground alignment,

    Z. Zha, H. Tang, Y . Sun, and J. Tang, “Boosting few-shot fine-grained recognition with background suppression and foreground alignment,” IEEE Transactions on Circuits and Systems for Video Technology, 2023

    2023

  21. [29]

    Fine-grained visual classification via internal ensemble learning transformer,

    Q. Xu, J. Wang, B. Jiang, and B. Luo, “Fine-grained visual classification via internal ensemble learning transformer,” IEEE Transactions on Multimedia, vol. 25, pp. 9015–9028, 2023

    2023

  22. [30]

    Delving into multi- modal prompting for fine-grained visual classification,

    X. Jiang, H. Tang, J. Gao, X. Du, S. He, and Z. Li, “Delving into multi- modal prompting for fine-grained visual classification,” in Proceedings of the AAAI Conference on Artificial Intelligence , 2024, pp. 2570–2578

    2024

  23. [31]

    Part-based r-cnns for fine-grained category detection,

    N. Zhang, J. Donahue, R. B. Girshick, and T. Darrell, “Part-based r-cnns for fine-grained category detection,” in Proceedings of the European Conference on Computer Vision , 2014, pp. 834–849

    2014

  24. [32]

    P-CNN: part-based con- volutional neural networks for fine-grained visual categorization,

    J. Han, X. Yao, G. Cheng, X. Feng, and D. Xu, “P-CNN: part-based con- volutional neural networks for fine-grained visual categorization,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 2, pp. 579–590, 2022

    2022

  25. [33]

    Cross-part learning for fine-grained image classification,

    M. Liu, C. Zhang, H. Bai, R. Zhang, and Y . Zhao, “Cross-part learning for fine-grained image classification,” IEEE Transactions on Image Processing, vol. 31, pp. 748–758, 2022

    2022

  26. [34]

    Imagdressing-v1: Customizable virtual dressing,

    F. Shen, X. Jiang, X. He, H. Ye, C. Wang, X. Du, Z. Li, and J. Tang, “Imagdressing-v1: Customizable virtual dressing,” arXiv preprint arXiv:2407.12705, 2024

    2024 arXiv

  27. [35]

    Hyperbolic vision transformers: Combining improvements in metric learning,

    A. Ermolov, L. Mirvakhabova, V . Khrulkov, N. Sebe, and I. V . Oseledets, “Hyperbolic vision transformers: Combining improvements in metric learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2022, pp. 7399–7409. JOURNAL OF LATEX CLASS ...

    2022

  28. [36]

    Proxynca++: Revisiting and revitalizing proxy neighborhood component analysis,

    E. W. Teh, T. DeVries, and G. W. Taylor, “Proxynca++: Revisiting and revitalizing proxy neighborhood component analysis,” in Proceedings of the European Conference on Computer Vision , 2020, pp. 448–464

    2020

  29. [37]

    Boosting vision transformers for image retrieval,

    C. H. Song, J. Yoon, S. Choi, and Y . Avrithis, “Boosting vision transformers for image retrieval,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2023, pp. 107– 117

    2023

  30. [38]

    Re-id- leak: Membership inference attacks against person re-identification,

    J. Gao, X. Jiang, S. Dou, D. Li, D. Miao, and C. Zhao, “Re-id- leak: Membership inference attacks against person re-identification,” International Journal of Computer Vision , pp. 1–15, 2024

    2024

  31. [39]

    Git: Graph interactive transformer for vehicle re-identification,

    F. Shen, Y . Xie, J. Zhu, X. Zhu, and H. Zeng, “Git: Graph interactive transformer for vehicle re-identification,” IEEE Transactions on Image Processing, vol. 32, pp. 1039–1051, 2023

    2023

  32. [40]

    Deep saliency hashing for fine-grained retrieval,

    S. Jin, H. Yao, X. Sun, S. Zhou, L. Zhang, and X. Hua, “Deep saliency hashing for fine-grained retrieval,” IEEE Transactions on Image Processing, vol. 29, pp. 5336–5351, 2020

    2020

  33. [41]

    A 2-net: Learning attribute-aware hash codes for large-scale fine-grained image retrieval,

    X. Wei, Y . Shen, X. Sun, H. Ye, and J. Yang, “A 2-net: Learning attribute-aware hash codes for large-scale fine-grained image retrieval,” in Proceedings of the Conference on Neural Information Processing Systems, 2021, pp. 5720–5730

    2021

  34. [42]

    Fine-grained hashing with double filtering,

    Z. Chen, X. Luo, Y . Wang, S. Guo, and X. Xu, “Fine-grained hashing with double filtering,” IEEE Transactions on Image Process. , vol. 31, pp. 1671–1683, 2022

    2022

  35. [43]

    Deep progressive asymmetric quantization based on causal intervention for fine-grained image retrieval,

    L. Ma, H. Hong, F. Meng, Q. Wu, and J. Wu, “Deep progressive asymmetric quantization based on causal intervention for fine-grained image retrieval,” IEEE Transactions on Multimedia , vol. 26, pp. 1306– 1318, 2024

    2024

  36. [44]

    Deep neighbor- hood structure-preserving hashing for large-scale image retrieval,

    Q. Qin, K. Xie, W. Zhang, C. Wang, and L. Huang, “Deep neighbor- hood structure-preserving hashing for large-scale image retrieval,” IEEE Transactions on Multimedia , vol. 26, pp. 1881–1893, 2024

    2024

  37. [45]

    Supervised deep hashing for scalable face image retrieval,

    J. Tang, Z. Li, and X. Zhu, “Supervised deep hashing for scalable face image retrieval,” Pattern Recognition, vol. 75, pp. 25–32, 2018

    2018

  38. [46]

    Weakly-supervised semantic guided hashing for social image retrieval,

    Z. Li, J. Tang, L. Zhang, and J. Yang, “Weakly-supervised semantic guided hashing for social image retrieval,” International Journal of Computer Vision, vol. 128, no. 8, pp. 2265–2278, 2020

    2020

  39. [47]

    Fast locality-sensitive hashing,

    A. Dasgupta, R. Kumar, and T. Sarl ´os, “Fast locality-sensitive hashing,” in Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , 2011, pp. 1073–1081

    2011

  40. [48]

    HHF: hashing-guided hinge function for deep hashing retrieval,

    C. Xu, Z. Chai, Z. Xu, H. Li, Q. Zuo, L. Yang, and C. Yuan, “HHF: hashing-guided hinge function for deep hashing retrieval,” IEEE Transactions on Multimedia , vol. 25, pp. 7428–7440, 2023

    2023

  41. [49]

    Iterative quantiza- tion: A procrustean approach to learning binary codes for large-scale image retrieval,

    Y . Gong, S. Lazebnik, A. Gordo, and F. Perronnin, “Iterative quantiza- tion: A procrustean approach to learning binary codes for large-scale image retrieval,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 12, pp. 2916–2929, 2013

    2013

  42. [50]

    Asymmetric deep supervised hashing,

    Q. Jiang and W. Li, “Asymmetric deep supervised hashing,” in Pro- ceedings of the AAAI Conference on Artificial Intelligence , 2018, pp. 3342–3349

    2018

  43. [51]

    Self-paced relational contrastive hash- ing for large-scale image retrieval,

    Z. Lu, L. Jin, Z. Li, and J. Tang, “Self-paced relational contrastive hash- ing for large-scale image retrieval,” IEEE Transactions on Multimedia , vol. 26, pp. 3392–3404, 2024

    2024

  44. [52]

    Alleviating over-fitting in hashing-based fine-grained image retrieval: From causal feature learning to binary-injected hash learning,

    X. Xiang, X. Ding, L. Jin, Z. Li, J. Tang, and R. Jain, “Alleviating over-fitting in hashing-based fine-grained image retrieval: From causal feature learning to binary-injected hash learning,” IEEE Transactions on Multimedia, pp. 1–13, 2024

    2024

  45. [53]

    Densifying one permutation hashing via rota- tion for fast near neighbor search,

    A. Shrivastava and P. Li, “Densifying one permutation hashing via rota- tion for fast near neighbor search,” in Proceedings of the International Conference on Machine Learning , 2014, pp. 557–565

    2014

  46. [54]

    Nvit: Vision transformer compression and parameter redistribution,

    H. Yang, H. Yin, P. Molchanov, H. Li, and J. Kautz, “Nvit: Vision transformer compression and parameter redistribution,” CoRR, vol. abs/2110.04869, 2021

    2021 arXiv

  47. [55]

    Chasing sparsity in vision transformers: An end-to-end exploration,

    T. Chen, Y . Cheng, Z. Gan, L. Yuan, L. Zhang, and Z. Wang, “Chasing sparsity in vision transformers: An end-to-end exploration,” in Proceed- ings of the Conference on Neural Information Processing Systems, 2021, pp. 19 974–19 988

    2021

  48. [56]

    Ia-red 2: Interpretability-aware redundancy reduction for vision transformers,

    B. Pan, Y . Jiang, R. Panda, Z. Wang, R. Feris, and A. Oliva, “Ia-red 2: Interpretability-aware redundancy reduction for vision transformers,” CoRR, vol. abs/2106.12620, 2021

    2021 arXiv

  49. [58]

    Evo-vit: Slow-fast token evolution for dynamic vision transformer,

    Y . Xu, Z. Zhang, M. Zhang, K. Sheng, K. Li, W. Dong, L. Zhang, C. Xu, and X. Sun, “Evo-vit: Slow-fast token evolution for dynamic vision transformer,” in Proceedings of the AAAI Conference on Artificial Intelligence, 2022, pp. 2964–2972

    2022

  50. [59]

    Evit: Expediting vision transformers via token reorganizations,

    Y . Liang, C. Ge, Z. Tong, Y . Song, J. Wang, and P. Xie, “Evit: Expediting vision transformers via token reorganizations,” in Proceedings of the International Conference on Learning Representations , 2022

    2022

  51. [60]

    Squeeze-and-excitation networks,

    J. Hu, L. Shen, and G. Sun, “Squeeze-and-excitation networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7132–7141

    2018

  52. [61]

    Vitkd: Feature- based knowledge distillation for vision transformers,

    Z. Yang, Z. Li, A. Zeng, Z. Li, C. Yuan, and Y . Li, “Vitkd: Feature- based knowledge distillation for vision transformers,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 1379–1388

    2024

  53. [62]

    Transformer- based distillation hash learning for image retrieval,

    Y . Lv, C. Wang, W. Yuan, X. Qian, W. Yang, and W. Zhao, “Transformer- based distillation hash learning for image retrieval,” Electronics, vol. 11, no. 18, p. 2810, 2022

    2022

  54. [63]

    Distilling the knowledge in a neural network,

    G. E. Hinton, O. Vinyals, and J. Dean, “Distilling the knowledge in a neural network,” CoRR, vol. abs/1503.02531, 2015

    2015 arXiv

  55. [64]

    Deep listwise triplet hashing for fine-grained image retrieval,

    Y . Liang, Y . Pan, H. Lai, W. Liu, and J. Yin, “Deep listwise triplet hashing for fine-grained image retrieval,” IEEE Transactions on Image Processing, vol. 31, pp. 949–961, 2022

    2022

  56. [65]

    Sub-region localized hashing for fine-grained image retrieval,

    X. Xiang, Y . Zhang, L. Jin, Z. Li, and J. Tang, “Sub-region localized hashing for fine-grained image retrieval,” IEEE Transactions on Image Processing, vol. 31, pp. 314–326, 2022

    2022

  57. [66]

    Exchnet: A unified hashing network for large-scale fine-grained image retrieval,

    Q. Cui, Q. Jiang, X. Wei, W. Li, and O. Yoshie, “Exchnet: A unified hashing network for large-scale fine-grained image retrieval,” in Pro- ceedings of the European Conference on Computer Vision , 2020, pp. 189–205

    2020

  58. [67]

    Building a bird recognition app and large scale dataset with citizen scientists: The fine print in fine-grained dataset collection,

    G. V . Horn, S. Branson, R. Farrell, S. Haber, J. Barry, P. Ipeirotis, P. Perona, and S. J. Belongie, “Building a bird recognition app and large scale dataset with citizen scientists: The fine print in fine-grained dataset collection,” in Proceedings of the IEEE Conference on ...

    2015

  59. [68]

    Vegfru: A domain-specific dataset for fine-grained visual categorization,

    S. Hou, Y . Feng, and Z. Wang, “Vegfru: A domain-specific dataset for fine-grained visual categorization,” in Proceedings of the IEEE International Conference on Computer Vision , 2017, pp. 541–549

    2017

  60. [69]

    Food-101 - mining discriminative components with random forests,

    L. Bossard, M. Guillaumin, and L. V . Gool, “Food-101 - mining discriminative components with random forests,” in Proceedings of the European Conference on Computer Vision , 2014, pp. 446–461

    2014

  61. [70]

    The inaturalist species classifi- cation and detection dataset,

    G. Van Horn, O. Mac Aodha, Y . Song, Y . Cui, C. Sun, A. Shepard, H. Adam, P. Perona, and S. Belongie, “The inaturalist species classifi- cation and detection dataset,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018, pp. 8769–8778

    2018

  62. [71]

    Char- acteristics matching based hash codes generation for efficient fine- grained image retrieval,

    Z.-D. Chen, L.-J. Zhao, Z.-C. Zhang, X. Luo, and X.-S. Xu, “Char- acteristics matching based hash codes generation for efficient fine- grained image retrieval,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 17 273–17 281

    2024

Pith tools

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