Pith. sign in

REVIEW 3 major objections 8 minor 99 references

Mr. DETR++: Instructive Multi-Route Training for Detection Transformers with Mixture-of-Experts

T0 review · 3 major / 8 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 Training detection transformers with two auxiliary one-to-many routes, discarded at inference, improves mAP by up to 4.0 points on COCO while leaving the deployed model unchanged.

desk verdict A genuinely useful multi-route training recipe with consistent gains, but the missing total-loss weighting equation and no code make it currently unreproducible; referee it and demand the fix. read the letter →

arxiv 2412.10028 v4 pith:H7ESHR5E submitted 2024-12-13 cs.CV

Chang-Bin Zhang , Yujie Zhong , Kai Han This is my paper · ORCID
classification cs.CV
keywords objectdetectiontransformersDETRone-to-manyassignmentmulti-routetraininginstructiveself-attentionmixture-of-expertspanopticsegmentation
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 asks what must be separated, inside a DETR-style transformer decoder, for one-to-one matching and one-to-many auxiliary supervision to coexist during training. Its empirical answer is that any single independent component — self-attention, cross-attention, or feed-forward network — suffices, even with all other components shared; fully sharing every component degrades the primary one-to-one prediction. From this it builds a three-route training scheme with a primary one-to-one route and two auxiliary one-to-many routes, later upgraded to a route-aware mixture-of-experts and an encoder MoE. The auxiliary routes are discarded at inference, so the deployed model keeps the baseline architecture and cost. Reported gains on COCO validation reach 4.0 mAP over Deformable-DETR++, 3.2 over DINO, and 2.0 over Align-DETR in 12-epoch schedules, with further gains on Objects365, NuImages, instance segmentation, and panoptic segmentation.

What carries the argument

The load-bearing machinery is the multi-route decoder. Route-2 is the primary one-to-one route and is identical to the baseline; Route-1 is an auxiliary one-to-many route with an independent feed-forward network; Route-3 is an auxiliary one-to-many route whose instructive self-attention concatenates $m$ learnable instruction tokens to the object queries, runs self-attention over the combined sequence, and then discards the token outputs, so the tokens condition the shared queries without surviving to inference. The one-to-many supervision uses the matching score $M_{ij} = \alpha s_i + (1-\alpha)\,\mathrm{IoU}(b_i,\bar{b}_j)$, keeping at most $K$ positive candidates above an IoU threshold $\tau$. In Mr. DETR++ the two FFN roles are replaced by a route-aware MoE: a pool of experts shared by all routes, with Route-1 gated independently and Routes 2 and 3 sharing a gate, using top-$k$ sparse activation. A scale-aware MoE applies gated experts only to low-scale encoder features while all tokens keep a shared FFN, and a localization-aware calibration $s_{\mathrm{calib}} = s_{\mathrm{cls}}^{\phi} s_{\mathrm{iou}}^{1-\phi}$ adjusts classification during inference.

What would settle it

Run the 12-epoch Deformable-DETR++ (300-query) setup with only the instructive self-attention route added: the paper's recipe predicts primary-route AP of 50.4, up from 47.6. A reader can check that single number; if it does not reproduce, the instructive self-attention is not carrying the reported gain.

Watch

Extended reading notes

Core claim

The central claim is that one-to-one and one-to-many objectives are not inherently in conflict inside a detection transformer; they become separable as soon as the decoder gives each objective its own copy of some minimal component. The paper demonstrates this with a systematic variant study: sharing everything drops the one-to-one route by 6.0 AP, whereas making only self-attention, only cross-attention, or only the FFN independent improves it by 2.1, 1.6, and 2.0 AP respectively. The resulting recipe, Mr. DETR++, runs three parallel routes on shared object queries and heads: a primary one-to-one route, an auxiliary route with an independent FFN, and an auxiliary route whose instructive self-attention attaches learnable instruction tokens to guide queries toward one-to-many targets. The two auxiliary routes are supervised by a one-to-many assignment and thrown away at deployment. An added route-aware MoE lets the two FFN variants share experts through separate gates, a scale-aware MoE strengthens low-resolution encoder features, and a localization-aware score calibrates classification at inference; the paper reports consistent mAP gains across baselines and tasks.

Load-bearing premise

The method's gains depend on one hand-set matching rule — up to six positive boxes per target, weighted by classification confidence and an IoU floor of 0.4 — that was tuned on COCO validation and is assumed to transfer to other datasets and tasks without re-tuning; if that rule is COCO-specific, the reported cross-domain improvements could shrink.

Editorial extensions

If this is right

  • Training-only upgrade: any DETR-family detector can take the three-route recipe without changing its inference graph; the paper verifies this on Deformable-DETR++, DINO, and Align-DETR.
  • The reported COCO gains are largest in short schedules: +4.0 AP at 12 epochs for Deformable-DETR++ (300 queries), +3.2 for DINO, and +2.0 for Align-DETR, with smaller but positive gains at 24 epochs.
  • Cross-domain gains follow the same recipe without per-dataset retuning: +4.5 AP on Objects365 and +3.8 AP on NuImages with Deformable-DETR++.
  • The same multi-route training lifts mask AP by 5.3 points on instance segmentation and PQ by 2.2 points on panoptic segmentation over the respective baselines.
  • The route-aware MoE visualisation shows Route-2 and the auxiliary routes selecting overlapping but distinct experts, which the paper reads as evidence that shared experts can specialise per route.

Reading between the lines

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

  • Editorial inference: if the 'any one independent component suffices' result is general, it suggests a cheap diagnostic for other multi-task transformers: duplicate only the component most coupled to each objective and share the rest, rather than building separate branches.
  • Editorial inference: instructive self-attention is essentially a soft-prompt mechanism on object queries; the same token-conditioning idea could steer one decoder toward different label assignments, segmentation heads, or multi-dataset objectives with no inference-time cost.
  • Editorial inference: the scale-aware encoder MoE ablation defines an accuracy-cost frontier (adding x32 and x16 scales costs about 16.6 GFLOPs for +0.1 AP), so the marginal value of more MoE scales is small and probably dataset-dependent.
  • Editorial note: the paper never writes the total loss built from the three routes, so an exact reproduction requires choosing route weights that the text does not specify.
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

3 major / 8 minor

Summary. The paper proposes a training-only enhancement for DETR-family detectors. It first treats an auxiliary one-to-many supervision branch as a second task and, through the ablations in Table I, concludes that sharing all decoder components between one-to-one and one-to-many targets degrades the primary route while making any single component independent helps. On this basis it introduces three training routes: Route-2, the primary one-to-one route; Route-1, an auxiliary one-to-many route with an independent FFN; and Route-3, an auxiliary one-to-many route with a novel instructive self-attention that concatenates learnable instruction tokens to the object queries. Mr. DETR++ extends this with a route-aware MoE in the decoder, a scale-aware encoder MoE, and a localization-aware score calibration. The auxiliary routes are discarded at inference, so the architecture and inference cost match the baseline. Experiments on COCO 2017, Objects365, and NuImages report consistent mAP improvements over Deformable-DETR++, DINO, and Align-DETR, with further results on instance and panoptic segmentation.

Significance. The empirical pattern in Table I is a useful and nontrivial contribution: it maps which decoder components can be shared between the two assignment targets, and it directly motivates the multi-route design. The proposed instruction tokens are an inexpensive way to specialize shared self-attention, and the route-aware MoE with separate gating for the primary route is a reasonable way to share experts while limiting route conflict. The evaluation is broad: three datasets, three baselines, two additional tasks, training-cost measurements, and a probing analysis of the mechanism. The paper also states the relation to the preliminary CVPR version [21] explicitly and credits prior work. If the missing training-objective details and single-run results are addressed, the work would be a solid empirical contribution to the DETR training literature. In its current form, the missing total-loss specification prevents the central claimed gains from being independently reproduced.

major comments (3)
  1. [III-B, Eq. (1)-(2), Table I] The total training objective is never written down. Eq. (1) defines the one-to-one loss for the bipartite match, and the text says the one-to-many losses are 'calculated as in Eqn. (1)' after Eq. (2), but no equation or sentence specifies how the Route-2 one-to-one loss and the Route-1/Route-3 one-to-many losses are combined, and no weighting coefficients are reported. This is load-bearing for a multi-task training method: with K=6 and alpha=0.3 the one-to-many assignment yields many more positives than the one-to-one match, so the relative loss weights determine whether the auxiliary gradients dominate or vanish. Table I already shows that the o2o/o2m balance is sensitive, since the 'Share All' configuration collapses the primary route from 47.6 to 41.6. Please add the explicit total-loss equation, report the route weights, and include an ablation over those weights.
  2. [IV-A, App. B-E, Tables II-VIII] All free hyperparameters are selected on the COCO 2017 validation set and then applied without re-tuning to Objects365, NuImages, and the segmentation tasks: K=6, alpha=0.3, tau=0.4 (App. D), 10 instruction tokens (App. B), 4 experts with top-2 (App. C), and phi=0 (App. E). Because all reported numbers are single runs without error bars, the claim of 'consistent improvements' across datasets and tasks is hard to evaluate: the cross-domain gains could be partly due to settings that are accidental to COCO-val. Please provide at least three seeds (or equivalent variance estimates) for the main comparisons and a sensitivity analysis of K, alpha, tau on at least one non-COCO benchmark.
  3. [IV, Tables II-XI] The manuscript does not state that code will be released, and the project page link does not provide a repository. For a method whose only evidence is empirical, the missing total-loss equation (Major 1) combined with absent code and single-run results is a substantial reproducibility gap. Please either release the code or provide a complete implementation specification, including the loss combination, route-wise loss masking, MoE auxiliary losses or their absence, and the exact training schedule for each dataset.
minor comments (8)
  1. [III-D, Eq. (13)] The definition after the equation uses 'spred' but the formula uses s_iou; align the notation.
  2. [III-B, Eq. (6), Fig. 4 caption] The typo 'InstrcutSA' should be 'InstructSA'.
  3. [IV-B.2 and Table IV caption] The text says Objects365 models are trained for 4 epochs, while the caption says 'All models are trained for 12 epochs'; correct the inconsistency and state the learning-rate schedule precisely.
  4. [II] The phrase 'parallel axillary queries' should read 'parallel auxiliary queries'.
  5. [App. B, Table XIIIa] The text says 'performance improves with an increased number of instruction tokens,' but the table shows 50.2, 50.1, 50.7, 50.4, and 50.5 for 1, 5, 10, 50, and 100 tokens, which is not monotonic; rephrase the conclusion.
  6. [III-D, Eq. (11)] Specify whether the gating scores c_i are normalized over the selected top-k experts and whether the MoE block includes a residual connection; also state whether any load-balancing loss is used for the route-aware MoE.
  7. [IV-F, Table IX] The table and text compare 'instruction by concatenation (ours)' against variants, but the baseline configuration (which routes are active) should be stated explicitly for each row so the reader can connect Table IX to Table VIII.
  8. [IV-A] Please add a reproducibility statement covering code availability, random seeds, and hardware details for each table.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the method is an empirical training scheme with self-contained ablations; the only self-citation is a pointer to a preliminary version and is not load-bearing.

full rationale

This paper does not derive a quantity from an input that already contains it. The central claims are empirical: Table I independently establishes that any single independent decoder component improves one-to-one prediction under auxiliary one-to-many training, and Tables II-VIII report mAP gains on COCO validation, Objects365, NuImages, and segmentation tasks. Hyperparameters K, alpha, and tau are tuned on COCO validation (App. D), but the reported method is not a 'prediction' of those values; it is a detector trained with the tuned assigner, so no fitted input is renamed as a prediction. The one-to-many loss is described as 'calculated as in Eqn. (1)' and the total multi-route loss weighting is never explicitly written, which is a reproducibility gap and a correctness risk, but it is not circularity because no equation is shown to reduce to its own input. The self-citation to the preliminary Mr. DETR paper [21] is an explicit provenance statement, and the current paper contains the actual experiments and ablations supporting the claims; the citation is not used to justify the result. No uniqueness theorem or ansatz is imported from the authors' prior work. The route-aware MoE, scale-aware MoE, and score calibration are new components tested against baselines, not renamed versions of their inputs. Therefore the derivation chain is self-contained and no circular step is exhibited.

Assumptions & free parameters 8 free parameters · 3 assumptions · 2 invented entities

The method contributes a training recipe; the free parameters are validation-set-tuned hyperparameters that the central claim depends on. No new physical entities are postulated; instruction tokens and gating designs are model components with only internal evidence.

free parameters (8)
  • K = 6
    Max positive candidates in one-to-many assignment (Eq. 2), chosen by ablation in App. D Fig. 11(a).
  • alpha = 0.3
    Weight of classification confidence in matching score M (Eq. 2), chosen by ablation in Fig. 11(b).
  • tau = 0.4
    IoU filter threshold in one-to-many assignment, chosen by ablation in Fig. 11(c).
  • num_instruction_tokens = 10 (per decoder layer)
    Number of learnable instruction tokens, set by ablation in App. B, Tab. XIIIa.
  • num_experts and top-k = 4 experts, top-2
    MoE configuration in encoder and decoder, chosen by ablation in App. C, Tab. XIV.
  • eta (MoE scales) = 1 (x64 downsampled features)
    Number of low-scale feature levels receiving encoder MoE (Eq. 12), chosen by ablation in Tab. X.
  • phi = 0 (default)
    Balance exponent in score calibration (Eq. 13); ablation in App. E shows AP insensitive but AP50/AP75 trade off.
  • starget exponent = 0.75
    Exponent for lifting the IoU target in the localization-aware score, set empirically in Sec. III-D.
assumptions (3)
  • domain assumption One-to-many assignment from DETA/DAC-DETR/MS-DETR provides a useful training signal.
    Adopted in Eq. (2) without proof; central to all auxiliary routes.
  • domain assumption Auxiliary routes can be discarded at inference without affecting the primary route.
    True by construction since routes are training-only branches; used to claim no inference cost.
  • domain assumption DETR decoder components are separable and can be independently assigned to routes.
    Underlies the design of Route-1 and Route-3 in Sec. III-B.
invented entities (2)
  • Learnable instruction tokens
    purpose: Attached to object queries in self-attention to guide one-to-many prediction
    Introduced in Sec. III-C; supported only by internal ablations (Tab. IX) and attention maps, no external falsifiable prediction.
  • Route-aware MoE gating
    purpose: Separate gating functions G and G' to share experts while avoiding route conflict
    Introduced in Sec. III-D; evidence is internal expert-distribution visualization (Fig. 9) and ablations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mr. DETR++: Instructive Multi-Route Training for Detection Transformers with Mixture-of-Experts." pith.science (2026). https://pith.science/paper/H7ESHR5E

@misc{pith2026241210028,
  author       = {Pith},
  title        = {Pith review of: Mr. DETR++: Instructive Multi-Route Training for Detection Transformers with Mixture-of-Experts},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H7ESHR5E}},
  note         = {Machine review of arXiv:2412.10028}
}
read the original abstract

Existing methods enhance the training of detection transformers by incorporating an auxiliary one-to-many assignment. In this work, we treat the model as a multi-task framework, simultaneously performing one-to-one and one-to-many predictions. We investigate the roles of each component in the transformer decoder across these two training targets, including self-attention, cross-attention, and feed-forward network. Our empirical results demonstrate that any independent component in the decoder can effectively learn both targets simultaneously, even when other components are shared. This finding leads us to propose a multi-route training mechanism, featuring a primary route for one-to-one prediction and two auxiliary training routes for one-to-many prediction. We propose a novel instructive self-attention mechanism, integrated into the first auxiliary route, which dynamically and flexibly guides object queries for one-to-many prediction. For the second auxiliary route, we introduce a route-aware Mixture-of-Experts (MoE) to facilitate knowledge sharing while mitigating potential conflicts between routes. Additionally, we apply an MoE to low-scale features in the encoder, optimizing the balance between efficiency and effectiveness. The auxiliary routes are discarded during inference. We conduct extensive experiments across various object detection baselines, achieving consistent improvements as demonstrated in Fig. 1. Our method is highly flexible and can be readily adapted to other tasks. To demonstrate its versatility, we conduct experiments on both instance segmentation and panoptic segmentation, further validating its effectiveness. Project page: https://visual-ai.github.io/mrdetr/

Figures

Figures reproduced from arXiv: 2412.10028 by the authors.

Figure 1
Figure 1. Improvements over different baseline models. The results are reported on the COCO 2017 validation in mean Average Precision (mAP). From left to right: Deformable￾DETR++ [11] with 300 queries, Deformable-DETR++ [11] with 900 queries, Deformable-DETR++ [11] trained for 24 epochs, DINO [9], DINO trained for 24 epochs, Align￾DETR [12] and Align-DETR [12] configured with 24 training epochs. query no object object query n… view at source ↗
Figure 2
Figure 2. One-to-one vs. one-to-many assignments. (a) ‘One-to￾one’ assigns each ground-truth box to a single predicted box, while predictions not associated with any object are supervised by ‘no object’ (background). (b) ‘One-to-many’ allows each ground-truth box to be paired with multiple predicted boxes. to-one assignment [9], [15], [16]. Specifically, DN-DETR [15], Group-DETR, and DINO [9] utilize multiple groups of parall… view at source ↗
Figure 3
Figure 3. Different configurations of the transformer decoder with auxiliary one-to-many training. ‘SA’: self-attention. ‘CA’: cross-attention. ‘FFN’: feed-forward network. ‘o2o’: one-to-one prediction. ‘o2m’: one-to-many prediction. TABLE I: The AP performance of different variants. Each variant includes a primary route for one-to-one prediction and several auxiliary routes for one-to-many prediction. ‘o2o’: the performance … view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Our proposed multi-route training method in Mr. DETR. It includes three training routes: Route-1, Route-2, and Route-3. All three routes share the same object queries and detection heads for classification and regression. Route-2 serves as the primary route for one-to-…
Figure 5
Figure 5. Figure 5: Various implementations of instructive self-attention. instruction tokens are incorporated into shared object queries through addition. This approach necessitates a fixed number of instruction tokens equivalent to the query count. Unlike the addition approach, our meth…
Figure 6
Figure 6. Figure 6: Illustration of the proposed Mr. DETR++ architecture. Our approach replaces the two independent FFNs in Mr. DETR with a route-aware MoE, enabling knowledge sharing and task-specific specialization across routes. Two distinct gating functions, G and G’, are employed: G …
Figure 7
Figure 7. Figure 7: Evaluation results of each epoch. We utilize the Deformable-DETR++ (300 queries) as the baseline model, which is trained for 12 and 24 epochs, respectively. head based on Deformable-DETR++ [11] utilizing 300 queries. For simplicity, both one-to-one and one-to-many assi…
Figure 8
Figure 8. Figure 8: Visualization of attention maps for instructive self-attention. We use Deformable-DETR++ with 300 object queries and 10 instruction tokens for this visualization. The first 10 tokens are instruction tokens. The vertical and horizontal axes represent the Query and Key, …
Figure 9
Figure 9. Figure 9: Distribution of Activated Experts in the Route-aware MoE. ‘Route-2’: the primary route for one-to-one prediction. ‘Route-3’: the auxiliary training route with instructive self￾attention. ‘Route-1’: the auxiliary training route with indepen￾dent gating. expert across al…
Figure 10
Figure 10. Figure 10: The cosine similarity between 10 instruction tokens used in Mr. DETR. TABLE XIV: Ablation study on configurations of MoE. (a) Configurations of MoE in transformer encoder experts top-k AP AP50 AP75 APs APm APl 2 1 51.7 68.8 56.5 35.3 55.3 66.1 4 1 51.9 68.9 56.6 35.0 …
Figure 11
Figure 11. Figure 11: Influence of hyper-parameters K, α and τ in the one-to-many assignment. (a) Influence of K for selecting top-K positive candidates. (b) Influence of α that denotes the weight of classification confidence when forming the matching score M. (c) Influence of τ that is us…
Figure 12
Figure 12. Figure 12: Qualitative results of our method. Left: prediction results. Right: ground truth [PITH_FULL_IMAGE:figures/full_fig_p017_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

99 extracted references · 68 canonical work pages

  1. [21]

    Mr. detr: Instructive multi-route training for detection transformers,

    C.-B. Zhang, Y . Zhong, and K. Han, “Mr. detr: Instructive multi-route training for detection transformers,” in IEEE Conf. Comput. Vis. Pattern Recog., 2025. 3

    work page 2025

  2. [1]

    End-to-end object detection with transformers,

    N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, and S. Zagoruyko, “End-to-end object detection with transformers,” in Eur. Conf. Comput. Vis., 2020. 1, 3, 4, 10

    2020

  3. [2]

    Dab-detr: Dynamic anchor boxes are better queries for detr,

    S. Liu, F. Li, H. Zhang, X. Yang, X. Qi, H. Su, J. Zhu, and L. Zhang, “Dab-detr: Dynamic anchor boxes are better queries for detr,” in Int. Conf. Learn. Represent., 2022. 1, 3

    2022

  4. [3]

    Conditional detr for fast training convergence,

    D. Meng, X. Chen, Z. Fan, G. Zeng, H. Li, Y . Yuan, L. Sun, and J. Wang, “Conditional detr for fast training convergence,” in Int. Conf. Comput. Vis., 2021. 1, 3

    2021

  5. [4]

    Anchor detr: Query design for transformer-based detector,

    Y . Wang, X. Zhang, T. Yang, and J. Sun, “Anchor detr: Query design for transformer-based detector,” in AAAI Conf. Artif. Intell. , 2022. 1, 3

    2022

  6. [5]

    Faster r-cnn: Towards real-time object detection with region proposal networks,

    S. Ren, K. He, R. Girshick, and J. Sun, “Faster r-cnn: Towards real-time object detection with region proposal networks,” IEEE Trans. Pattern Anal. Mach. Intell. , 2016. 1, 3, 4

    2016

  7. [6]

    Focal loss for dense object detection,

    T.-Y . Lin, P. Goyal, R. B. Girshick, K. He, and P. Doll ´ar, “Focal loss for dense object detection,” Int. Conf. Comput. Vis. , 2017. 1, 3, 4

    2017

  8. [7]

    Fcos: A simple and strong anchor-free object detector,

    Z. Tian, C. Shen, H. Chen, and T. He, “Fcos: A simple and strong anchor-free object detector,” IEEE Trans. Pattern Anal. Mach. Intell. ,

Show all 99 references
  1. [8]

    Detrs with hybrid matching,

    D. Jia, Y . Yuan, H. He, X. Wu, H. Yu, W. Lin, L. Sun, C. Zhang, and H. Hu, “Detrs with hybrid matching,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023. 1, 3, 4, 7, 8, 9, 10, 12

    2023

  2. [9]

    Dino: Detr with improved denoising anchor boxes for end-to-end object detection,

    H. Zhang, F. Li, S. Liu, L. Zhang, H. Su, J. Zhu, L. Ni, and H.-Y . Shum, “Dino: Detr with improved denoising anchor boxes for end-to-end object detection,” in Int. Conf. Learn. Represent. , 2023. 1, 4, 7, 8, 9

    2023

  3. [10]

    Rethinking transformer- based set prediction for object detection,

    Z. Sun, S. Cao, Y . Yang, and K. M. Kitani, “Rethinking transformer- based set prediction for object detection,” in Int. Conf. Comput. Vis. ,

  4. [11]

    Deformable detr: Deformable transformers for end-to-end object detection,

    X. Zhu, W. Su, L. Lu, B. Li, X. Wang, and J. Dai, “Deformable detr: Deformable transformers for end-to-end object detection,” in Int. Conf. Learn. Represent., 2021. 1, 3, 7, 8, 9, 10, 12, 16, 17

    2021

  5. [12]

    Align-detr: Improving detr with simple iou-aware bce loss,

    Z. Cai, S. Liu, G. Wang, Z. Ge, X. Zhang, and D. Huang, “Align-detr: Improving detr with simple iou-aware bce loss,” in Brit. Mach. Vis. Conf.,

  6. [13]

    Dac-detr: Divide the attention layers and conquer,

    Z. Hu, Y . Sun, J. Wang, and Y . Yang, “Dac-detr: Divide the attention layers and conquer,” Adv. Neural Inform. Process. Syst. , 2024. 1, 4, 7, 8, 9, 10, 16

    2024

  7. [14]

    Ms-detr: Efficient detr training with mixed supervision,

    C. Zhao, Y . Sun, W. Wang, Q. Chen, E. Ding, Y . Yang, and J. Wang, “Ms-detr: Efficient detr training with mixed supervision,” in IEEE Conf. Comput. Vis. Pattern Recog., 2024. 1, 4, 7, 8, 9, 10, 12, 16

    2024

  8. [15]

    Dn-detr: Accelerate detr training by introducing query denoising,

    F. Li, H. Zhang, S. Liu, J. Guo, L. M. Ni, and L. Zhang, “Dn-detr: Accelerate detr training by introducing query denoising,” IEEE Trans. Pattern Anal. Mach. Intell. , 2024. 1, 4

    2024

  9. [16]

    Group detr: Fast detr training with group-wise one-to-many assignment,

    Q. Chen, X. Chen, J. Wang, S. Zhang, K. Yao, H. Feng, J. Han, E. Ding, G. Zeng, and J. Wang, “Group detr: Fast detr training with group-wise one-to-many assignment,” in Int. Conf. Comput. Vis. , 2023. 1, 4, 8

    2023

  10. [17]

    Nms strikes back,

    J. Ouyang-Zhang, J. H. Cho, X. Zhou, and P. Kr ¨ahenb¨uhl, “Nms strikes back,” arXiv preprint arXiv:2212.06137 , 2022. 1, 4, 16

    2022 arXiv

  11. [18]

    Varifocalnet: An iou-aware dense object detector,

    H. Zhang, Y . Wang, F. Dayoub, and N. Sunderhauf, “Varifocalnet: An iou-aware dense object detector,” in IEEE Conf. Comput. Vis. Pattern Recog., 2021. 3, 7

    2021

  12. [19]

    Detection transformer with stable matching,

    S. Liu, T. Ren, J. Chen, Z. Zeng, H. Zhang, F. Li, H. Li, J. Huang, H. Su, J. Zhu et al., “Detection transformer with stable matching,” in Int. Conf. Comput. Vis., 2023. 3, 7, 8, 9

    2023

  13. [20]

    Rank-detr for high quality object detection,

    Y . Pu, W. Liang, Y . Hao, Y . Yuan, Y . Yang, C. Zhang, H. Hu, and G. Huang, “Rank-detr for high quality object detection,” Adv. Neural Inform. Process. Syst. , 2024. 3, 7, 8, 9

    2024

  14. [22]

    Objects365: A large-scale, high-quality dataset for object detection,

    S. Shao, Z. Li, T. Zhang, C. Peng, G. Yu, X. Zhang, J. Li, and J. Sun, “Objects365: A large-scale, high-quality dataset for object detection,” in Int. Conf. Comput. Vis. , 2019. 3, 7, 9, 15

    2019

  15. [23]

    nuscenes: A multimodal dataset for autonomous driving,

    H. Caesar, V . Bankiti, A. H. Lang, S. V ora, V . E. Liong, Q. Xu, A. Krishnan, Y . Pan, G. Baldan, and O. Beijbom, “nuscenes: A multimodal dataset for autonomous driving,” in IEEE Conf. Comput. Vis. Pattern Recog., 2020. 3, 7, 9, 15

    2020

  16. [24]

    Yolox: Exceeding yolo series in 2021,

    G. Zheng, L. Songtao, W. Feng, L. Zeming, S. Jian et al. , “Yolox: Exceeding yolo series in 2021,” arXiv preprint arXiv:2107.08430 , 2021. 3

    2021 arXiv

  17. [25]

    You should look at all objects,

    Z. Jin, D. Yu, L. Song, Z. Yuan, and L. Yu, “You should look at all objects,” in Eur. Conf. Comput. Vis., 2022. 3

    2022

  18. [26]

    Dynamic detr: End-to-end object detection with dynamic attention,

    X. Dai, Y . Chen, J. Yang, P. Zhang, L. Yuan, and L. Zhang, “Dynamic detr: End-to-end object detection with dynamic attention,” in Int. Conf. Comput. Vis., 2021. 3

    2021

  19. [27]

    Fast convergence of detr with spatially modulated co-attention,

    P. Gao, M. Zheng, X. Wang, J. Dai, and H. Li, “Fast convergence of detr with spatially modulated co-attention,” in Int. Conf. Comput. Vis. ,

  20. [28]

    Cascade-detr: delving into high-quality universal object detection,

    M. Ye, L. Ke, S. Li, Y .-W. Tai, C.-K. Tang, M. Danelljan, and F. Yu, “Cascade-detr: delving into high-quality universal object detection,” in Int. Conf. Comput. Vis. , 2023. 3, 7, 8

    2023

  21. [29]

    Ease-detr: Easing the competition among object queries,

    Y . Gao, Y . Sun, X. Ding, C. Zhao, and S. Liu, “Ease-detr: Easing the competition among object queries,” in IEEE Conf. Comput. Vis. Pattern Recog., 2024. 3, 8, 9

    2024

  22. [30]

    Relation detr: Exploring explicit position relation prior for object detection,

    X. Hou, M. Liu, S. Zhang, P. Wei, B. Chen, and X. Lan, “Relation detr: Exploring explicit position relation prior for object detection,” in Eur. Conf. Comput. Vis., 2024. 3, 9

    2024

  23. [31]

    Sap-detr: bridging the gap between salient points and queries-based transformer detector for fast model convergency,

    Y . Liu, Y . Zhang, Y . Wang, Y . Zhang, J. Tian, Z. Shi, J. Fan, and Z. He, “Sap-detr: bridging the gap between salient points and queries-based transformer detector for fast model convergency,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023. 3

    2023

  24. [32]

    Accelerating detr convergence via semantic-aligned matching,

    G. Zhang, Z. Luo, Y . Yu, K. Cui, and S. Lu, “Accelerating detr convergence via semantic-aligned matching,” in IEEE Conf. Comput. Vis. Pattern Recog., 2022. 3

    2022

  25. [33]

    Hybrid proposal refiner: Revisiting detr series from the faster r-cnn perspective,

    J. Zhao, F. Wei, and C. Xu, “Hybrid proposal refiner: Revisiting detr series from the faster r-cnn perspective,” in IEEE Conf. Comput. Vis. Pattern Recog., 2024. 3

    2024

  26. [34]

    Mi-detr: An object detection model with multi-time inquiries mechanism,

    Z. Nan, X. Li, J. Dai, and T. Xiang, “Mi-detr: An object detection model with multi-time inquiries mechanism,” arXiv preprint arXiv:2503.01463 ,

    arXiv

  27. [35]

    Efficient detr: improving end-to-end object detector with dense prior,

    Z. Yao, J. Ai, B. Li, and C. Zhang, “Efficient detr: improving end-to-end object detector with dense prior,” arXiv preprint arXiv:2104.01318, 2021. 3

    2021 arXiv

  28. [36]

    Salience detr: Enhancing detection transformer with hierarchical salience filtering refinement,

    X. Hou, M. Liu, S. Zhang, P. Wei, and B. Chen, “Salience detr: Enhancing detection transformer with hierarchical salience filtering refinement,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2024. 3, 8, 9

    2024

  29. [37]

    Dense distinct query for end-to-end object detection,

    S. Zhang, X. Wang, J. Wang, J. Pang, C. Lyu, W. Zhang, P. Luo, and K. Chen, “Dense distinct query for end-to-end object detection,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2023. 3, 8

    2023

  30. [38]

    Sparse detr: Efficient end-to-end object detection with learnable sparsity,

    B. Roh, J. Shin, W. Shin, and S. Kim, “Sparse detr: Efficient end-to-end object detection with learnable sparsity,” in Int. Conf. Learn. Represent. ,

  31. [39]

    Dˆ 2etr: Decoder- only detr with computationally efficient cross-scale attention,

    J. Lin, X. Mao, Y . Chen, L. Xu, Y . He, and H. Xue, “Dˆ 2etr: Decoder- only detr with computationally efficient cross-scale attention,” arXiv preprint arXiv:2203.00860, 2022. 3

    2022 arXiv

  32. [40]

    Detrs beat yolos on real-time object detection,

    Y . Zhao, W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y . Liu, and J. Chen, “Detrs beat yolos on real-time object detection,” in IEEE Conf. Comput. Vis. Pattern Recog., 2024. 3

    2024

  33. [41]

    Lite detr: An interleaved multi-scale encoder for efficient detr,

    F. Li, A. Zeng, S. Liu, H. Zhang, H. Li, L. Zhang, and L. M. Ni, “Lite detr: An interleaved multi-scale encoder for efficient detr,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023. 3

    2023

  34. [42]

    Lw-detr: A transformer replacement to yolo for real-time detection,

    Q. Chen, X. Su, X. Zhang, J. Wang, J. Chen, Y . Shen, C. Han, Z. Chen, W. Xu, F. Li et al. , “Lw-detr: A transformer replacement to yolo for real-time detection,” arXiv preprint arXiv:2406.03459 , 2024. 3

    2024 arXiv

  35. [43]

    Decoupled detr: Spatially disentangling localization and classification for improved end-to-end object detection,

    M. Zhang, G. Song, Y . Liu, and H. Li, “Decoupled detr: Spatially disentangling localization and classification for improved end-to-end object detection,” in Int. Conf. Comput. Vis. , 2023. 3

    2023

  36. [44]

    Bridging the gap between anchor-based and anchor-free detection via adaptive training sample selection,

    S. Zhang, C. Chi, Y . Yao, Z. Lei, and S. Z. Li, “Bridging the gap between anchor-based and anchor-free detection via adaptive training sample selection,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2020. 3

    2020

  37. [45]

    Ota: Optimal transport assignment for object detection,

    Z. Ge, S. Liu, Z. Li, O. Yoshie, and J. Sun, “Ota: Optimal transport assignment for object detection,” in IEEE Conf. Comput. Vis. Pattern Recog., 2021. 3

    2021

  38. [46]

    Tood: Task- aligned one-stage object detection,

    C. Feng, Y . Zhong, Y . Gao, M. R. Scott, and W. Huang, “Tood: Task- aligned one-stage object detection,” in Int. Conf. Comput. Vis. , 2021. 3

    2021

  39. [47]

    Adamixer: A fast-converging query-based object detector,

    Z. Gao, L. Wang, B. Han, and S. Guo, “Adamixer: A fast-converging query-based object detector,” in IEEE Conf. Comput. Vis. Pattern Recog. ,

  40. [48]

    Recurrent glimpse-based decoder for detection with transformer,

    Z. Chen, J. Zhang, and D. Tao, “Recurrent glimpse-based decoder for detection with transformer,” in IEEE Conf. Comput. Vis. Pattern Recog. ,

  41. [49]

    Querydet: Cascaded sparse query for accelerating high-resolution small object detection,

    C. Yang, Z. Huang, and N. Wang, “Querydet: Cascaded sparse query for accelerating high-resolution small object detection,” in IEEE Conf. Comput. Vis. Pattern Recog., 2022. 3

    2022

  42. [50]

    Exploring plain vision transformer backbones for object detection,

    Y . Li, H. Mao, R. Girshick, and K. He, “Exploring plain vision transformer backbones for object detection,” in Eur. Conf. Comput. Vis.,

  43. [51]

    Cf-detr: Coarse-to-fine transformers for end-to-end object detection,

    X. Cao, P. Yuan, B. Feng, and K. Niu, “Cf-detr: Coarse-to-fine transformers for end-to-end object detection,” in AAAI Conf. Artif. Intell. ,

  44. [52]

    Feataug-detr: Enriching one-to-many matching for detrs with feature augmentation,

    R. Fang, P. Gao, A. Zhou, Y . Cai, S. Liu, J. Dai, and H. Li, “Feataug-detr: Enriching one-to-many matching for detrs with feature augmentation,” IEEE Trans. Pattern Anal. Mach. Intell. , 2024. 3

    2024

  45. [53]

    The hungarian method for the assignment problem,

    H. W. Kuhn, “The hungarian method for the assignment problem,” Naval Res. Logist., 1955. 4

    1955

  46. [54]

    Learning dynamic query combinations for transformer-based object detection and segmentation,

    Y . Cui, L. Yang, and H. Yu, “Learning dynamic query combinations for transformer-based object detection and segmentation,” in Int. Conf. Mach. Learn., 2023. 4

    2023

  47. [55]

    Stageinteractor: Query-based object detector with cross-stage interaction,

    Y . Teng, H. Liu, S. Guo, and L. Wang, “Stageinteractor: Query-based object detector with cross-stage interaction,” in Int. Conf. Comput. Vis. ,

  48. [56]

    Enhanced training of query-based object detection via selective query recollection,

    F. Chen, H. Zhang, K. Hu, Y .-K. Huang, C. Zhu, and M. Savvides, “Enhanced training of query-based object detection via selective query recollection,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2023. 4

    2023

  49. [57]

    Detrs with collaborative hybrid assignments training,

    Z. Zong, G. Song, and Y . Liu, “Detrs with collaborative hybrid assignments training,” in Int. Conf. Comput. Vis. , 2023. 4, 9

    2023

  50. [58]

    Kd-detr: Knowledge distillation for detection transformer with consistent distillation points sampling,

    Y . Wang, X. Li, S. Weng, G. Zhang, H. Yue, H. Feng, J. Han, and E. Ding, “Kd-detr: Knowledge distillation for detection transformer with consistent distillation points sampling,” in IEEE Conf. Comput. Vis. Pattern Recog. ,

  51. [59]

    Detrdistill: A universal knowledge distillation framework for detr-families,

    J. Chang, S. Wang, H.-M. Xu, Z. Chen, C. Yang, and F. Zhao, “Detrdistill: A universal knowledge distillation framework for detr-families,” in Int. Conf. Comput. Vis., 2023. 4

    2023

  52. [60]

    Teach- detr: Better training detr with teachers,

    L. Huang, K. Lu, G. Song, L. Wang, S. Liu, Y . Liu, and H. Li, “Teach- detr: Better training detr with teachers,” IEEE Trans. Pattern Anal. Mach. Intell., 2023. 4

    2023

  53. [61]

    Distilling the knowledge in a neural network,

    G. Hinton, O. Vinyals, and J. Dean, “Distilling the knowledge in a neural network,” arXiv preprint arXiv:1503.02531 , 2015. 4

    2015 arXiv

  54. [62]

    Delving deep into label smoothing,

    C.-B. Zhang, P.-T. Jiang, Q. Hou, Y . Wei, Q. Han, Z. Li, and M.-M. Cheng, “Delving deep into label smoothing,” IEEE Trans. Image Process.,

  55. [63]

    Adaptive mixtures of local experts,

    R. A. Jacobs, M. I. Jordan, S. J. Nowlan, and G. E. Hinton, “Adaptive mixtures of local experts,” Neural computation, 1991. 4

    1991

  56. [64]

    Outrageously large neural networks: The sparsely-gated mixture-of-experts layer,

    N. Shazeer, A. Mirhoseini, K. Maziarz, A. Davis, Q. Le, G. Hinton, and J. Dean, “Outrageously large neural networks: The sparsely-gated mixture-of-experts layer,” in Int. Conf. Learn. Represent. , 2017. 4

    2017

  57. [65]

    Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity,

    W. Fedus, B. Zoph, and N. Shazeer, “Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity,” J. Machine Learn. Research, 2022. 4

    2022

  58. [66]

    Deepseek-v3 technical report,

    A. Liu, B. Feng, B. Xue, B. Wang, B. Wu, C. Lu, C. Zhao, C. Deng, C. Zhang, C. Ruan et al., “Deepseek-v3 technical report,” arXiv preprint arXiv:2412.19437, 2024. 4

    2024 arXiv

  59. [67]

    Scaling vision with sparse mixture of experts,

    C. Riquelme, J. Puigcerver, B. Mustafa, M. Neumann, R. Jenatton, A. Susano Pinto, D. Keysers, and N. Houlsby, “Scaling vision with sparse mixture of experts,” in Adv. Neural Inform. Process. Syst. , 2021. 4

    2021

  60. [68]

    Learning to merge tokens in vision transformers,

    C. Renggli, A. S. Pinto, N. Houlsby, B. Mustafa, J. Puigcerver, and C. Riquelme, “Learning to merge tokens in vision transformers,” arXiv preprint arXiv:2202.12015, 2022. 4

    2022 arXiv

  61. [69]

    Residual mixture of experts,

    L. Wu, M. Liu, Y . Chen, D. Chen, X. Dai, and L. Yuan, “Residual mixture of experts,” arXiv preprint arXiv:2204.09636 , 2022. 4

    2022 arXiv

  62. [70]

    Robust mixture-of-expert training for convolutional neural networks,

    Y . Zhang, R. Cai, T. Chen, G. Zhang, H. Zhang, P.-Y . Chen, S. Chang, Z. Wang, and S. Liu, “Robust mixture-of-expert training for convolutional neural networks,” in Int. Conf. Comput. Vis. , 2023. 4

    2023

  63. [71]

    Scaling diffusion transformers to 16 billion parameters,

    Z. Fei, M. Fan, C. Yu, D. Li, and J. Huang, “Scaling diffusion transformers to 16 billion parameters,” arXiv preprint arXiv:2407.11633 , 2024. 4

    2024 arXiv

  64. [72]

    M 3vit: Mixture-of-experts vision transformer for efficient multi- task learning with model-accelerator co-design,

    Z. Fan, R. Sarkar, Z. Jiang, T. Chen, K. Zou, Y . Cheng, C. Hao, Z. Wang et al., “M 3vit: Mixture-of-experts vision transformer for efficient multi- task learning with model-accelerator co-design,” in Adv. Neural Inform. Process. Syst., 2022. 4

    2022

  65. [73]

    Mod-squad: Designing mixtures of experts as modular multi-task learners,

    Z. Chen, Y . Shen, M. Ding, Z. Chen, H. Zhao, E. G. Learned-Miller, and C. Gan, “Mod-squad: Designing mixtures of experts as modular multi-task learners,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2023. 4

    2023

  66. [74]

    Adamv-moe: Adaptive multi-task vision mixture-of-experts,

    T. Chen, X. Chen, X. Du, A. Rashwan, F. Yang, H. Chen, Z. Wang, and Y . Li, “Adamv-moe: Adaptive multi-task vision mixture-of-experts,” in Int. Conf. Comput. Vis. , 2023. 4

    2023

  67. [75]

    Multi-task dense prediction via mixture of low-rank experts,

    Y . Yang, P.-T. Jiang, Q. Hou, H. Zhang, J. Chen, and B. Li, “Multi-task dense prediction via mixture of low-rank experts,” in IEEE Conf. Comput. Vis. Pattern Recog., 2024. 4

    2024

  68. [76]

    Lora: Low-rank adaptation of large language models,

    E. J. Hu, Y . Shen, P. Wallis, Z. Allen-Zhu, Y . Li, S. Wang, L. Wang, W. Chen et al., “Lora: Low-rank adaptation of large language models,” ICLR, 2022. 4

    2022

  69. [77]

    Language models are few-shot learners,

    B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal et al., “Language models are few-shot learners,” arXiv preprint arXiv:2005.14165 , 2020. 6

    2005 arXiv

  70. [78]

    Auto- prompt: Eliciting knowledge from language models with automatically generated prompts,

    T. Shin, Y . Razeghi, R. L. Logan IV , E. Wallace, and S. Singh, “Auto- prompt: Eliciting knowledge from language models with automatically generated prompts,” arXiv preprint arXiv:2010.15980 , 2020. 6

    2010 arXiv

  71. [79]

    How can we know what language models know?

    Z. Jiang, F. F. Xu, J. Araki, and G. Neubig, “How can we know what language models know?” Trans. Assoc. Comput. Linguist. , 2020. 6

    2020

  72. [80]

    Prefix-tuning: Optimizing continuous prompts for generation,

    X. L. Li and P. Liang, “Prefix-tuning: Optimizing continuous prompts for generation,” in Annu. Meet. Assoc. Comput. Linguist. , 2021. 6

    2021

  73. [81]

    The power of scale for parameter- efficient prompt tuning,

    B. Lester, R. Al-Rfou, and N. Constant, “The power of scale for parameter- efficient prompt tuning,” in Conf. Empir. Methods Nat. Lang. Process. ,

  74. [82]

    P-tuning v2: Prompt tuning can be comparable to fine-tuning universally across scales and tasks,

    X. Liu, K. Ji, Y . Fu, W. L. Tam, Z. Du, Z. Yang, and J. Tang, “P-tuning v2: Prompt tuning can be comparable to fine-tuning universally across scales and tasks,” in Annu. Meet. Assoc. Comput. Linguist. , 2022. 6

    2022

  75. [83]

    Learning to prompt for vision-language models,

    K. Zhou, J. Yang, C. C. Loy, and Z. Liu, “Learning to prompt for vision-language models,” Int. J. Comput. Vis. , 2022. 6

    2022

  76. [84]

    Prompting visual- language models for efficient video understanding,

    C. Ju, T. Han, K. Zheng, Y . Zhang, and W. Xie, “Prompting visual- language models for efficient video understanding,” in Eur. Conf. Comput. Vis., 2022. 6

    2022

  77. [85]

    Learning to prompt for continual learning,

    Z. Wang, Z. Zhang, C.-Y . Lee, H. Zhang, R. Sun, X. Ren, G. Su, V . Perot, J. Dy, and T. Pfister, “Learning to prompt for continual learning,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2022. 6

    2022

  78. [86]

    imove: Instance-motion-aware video understanding,

    J. Li, Y . Shi, Z. Ma, H. Xu, F. Cheng, H. Xiao, R. Kang, F. Yang, T. Gao, and D. Zhang, “imove: Instance-motion-aware video understanding,” arXiv preprint arXiv:2502.11594 , 2025. 6

    2025 arXiv

  79. [87]

    Visual prompt tuning,

    M. Jia, L. Tang, B.-C. Chen, C. Cardie, S. Belongie, B. Hariharan, and S.-N. Lim, “Visual prompt tuning,” in Eur. Conf. Comput. Vis., 2022. 6

    2022

  80. [88]

    Multitask vision-language prompt tuning,

    S. Shen, S. Yang, T. Zhang, B. Zhai, J. E. Gonzalez, K. Keutzer, and T. Darrell, “Multitask vision-language prompt tuning,” in IEEE Winter Conf. Appl. Comput. Vis. , 2024. 6

    2024

  81. [89]

    Improving visual prompt tuning for self-supervised vision transformers,

    S. Yoo, E. Kim, D. Jung, J. Lee, and S. Yoon, “Improving visual prompt tuning for self-supervised vision transformers,” in Int. Conf. Mach. Learn.,

  82. [90]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Eur. Conf. Comput. Vis., 2014. 7, 8, 9, 10, 15

    2014

  83. [91]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2016. 7, 8, 9, 10

    2016

  84. [92]

    Panoptic segmentation,

    A. Kirillov, K. He, R. Girshick, C. Rother, and P. Doll ´ar, “Panoptic segmentation,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2019. 7

    2019

  85. [93]

    Swin transformer: Hierarchical vision transformer using shifted windows,

    Z. Liu, Y . Lin, Y . Cao, H. Hu, Y . Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted windows,” in Int. Conf. Comput. Vis. , 2021. 7, 9, 16

    2021

  86. [94]

    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 IEEE Conf. Comput. Vis. Pattern Recog., 2009. 7

    2009

  87. [95]

    Decoupled weight decay regularization,

    I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” in Int. Conf. Learn. Represent. , 2017. 7

    2017

  88. [96]

    v-clr: View-consistent learning for open-world instance segmentation,

    C.-B. Zhang, J. Ni, Y . Zhong, and K. Han, “v-clr: View-consistent learning for open-world instance segmentation,” in IEEE Conf. Comput. Vis. Pattern Recog., 2025. 9

    2025

  89. [97]

    Mask dino: Towards a unified transformer-based framework for object detection and segmentation,

    F. Li, H. Zhang, H. Xu, S. Liu, L. Zhang, L. M. Ni, and H.-Y . Shum, “Mask dino: Towards a unified transformer-based framework for object detection and segmentation,” in IEEE Conf. Comput. Vis. Pattern Recog. ,

  90. [98]

    Masked- attention mask transformer for universal image segmentation,

    B. Cheng, I. Misra, A. G. Schwing, A. Kirillov, and R. Girdhar, “Masked- attention mask transformer for universal image segmentation,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2022. 10

    2022

  91. [99]

    Panoptic segformer: Delving deeper into panoptic segmentation with transformers,

    Z. Li, W. Wang, E. Xie, Z. Yu, A. Anandkumar, J. M. Alvarez, P. Luo, and T. Lu, “Panoptic segformer: Delving deeper into panoptic segmentation with transformers,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2022. 10 15 APPENDIX A. Datasets MS COCO 2017. This dataset [90] consi...

    2022

Pith tools

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