Pith. sign in

REVIEW 2 major objections 6 minor 72 references

MAFE R-CNN: Selecting More Samples to Learn Category-aware Features for Small Object Detection

T0 review · 2 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper argues that small-object detection fails because training samples are imbalanced and features are blurred, and that a size-aware sample selector plus a category-aware memory fixes both well enough to set new records on SODA-D…

desk verdict Incremental but well-ablated small-object detector with a shaky unverified normality assumption in its sample-selection mechanism; send to review but push for code, variance, and a softer claim. read the letter →

arxiv 2505.16442 v1 pith:XVRKKRYV submitted 2025-05-22 cs.CV

Yichen Li , Qiankun Liu , Zhenchao Jin , Jiuzhe Wei , Jing Nie , Ying Fu This is my paper · ORCID
classification cs.CV
keywords smallobjectdetectionsampleassignmentcategory-awarememorymulti-clueconfidencedynamicthresholdfeatureenhancementSODAbenchmarktwo-stagedetector
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

Small objects are hard to detect because fixed sample-assignment thresholds leave them with too few positive training samples, and because their few pixels yield features too noisy to discriminate. This paper argues that both defects can be repaired inside a standard two-stage detector, without extra supervision or auxiliary losses. Its Multi-Clue Sample Selection combines IoU overlap, predicted category confidence, and ground-truth size into one score, then computes a per-object dynamic threshold so that objects of every size receive a balanced number of positives. Its Category-aware Feature Enhancement Mechanism stores one prototype per category in a memory updated from ground-truth features, and lets each candidate box pool these prototypes by its predicted probabilities before a cross-attention step. On the SODA-D driving benchmark the resulting MAFE R-CNN reaches 32.7% AP and 46.1% AR, and on SODA-A it reaches 35.8% AP and 44.7% AR, the best reported numbers among the compared detectors.

What carries the argument

The load-bearing machinery is a pair of training-time mechanisms. MCSS computes for each ground truth $G_g$ the multi-clue confidence $D=\alpha D_c+(1-\alpha)D_{iou}$ over the $k$ nearest candidate boxes, where $D_c$ is the predicted category confidence and $D_{iou}$ is the IoU distance, and accepts a candidate when $D\ge T_g=\min(m_D+\gamma v_D,\beta)$; here $m_D$ and $v_D$ are the mean and standard deviation of $D$, and $\gamma=\min(S_g/S_{\max},3)$ scales with the object's absolute size. The normal-tail logic behind $m_D+\gamma v_D$ is what balances positive-sample counts across sizes. CFEM maintains a category-aware memory $M\in\mathbb{R}^{C\times D}$, updated by momentum averaging of ground-truth features weighted by inverse cosine similarity, generates for each candidate box a category-aware feature $F_c=P\cdot M$ from its predicted category probabilities $P$, and then enriches the box feature $R$ by cross-attention with $F_c$, producing $R_{\text{Enh}}=\text{Linear}(\text{Concat}(R,A(R,F_c)))$. The memory is updated only during training and frozen at inference.

What would settle it

During training on SODA-D, record the empirical distribution of multi-clue confidence scores $D$ for the candidate boxes of each ground truth in the second RoI head. For each object-size band, compare the fraction of candidates above $T_g=\min(m_D+\gamma v_D,\beta)$ with the fraction a normal distribution would predict; a systematic deviation such as near-zero positives for the smallest band would falsify the balancing mechanism even if final AP stays high.

Watch

Extended reading notes

Core claim

The central claim is that the detector's weakness on small objects is not an unavoidable resolution limit but a training-time problem of sample supply and feature quality. The paper shows that replacing fixed IoU or distance thresholds with the MCSS criterion -- a weighted sum of category confidence and IoU distance, with a threshold $T_g=\min(m_D+\gamma v_D,\beta)$ that depends on ground-truth size through $\gamma=\min(S_g/S_{\max},3)$ -- raises AP from 31.2 to 32.2 on SODA-D when used alone in the cascade. Adding CFEM alone raises AP to 32.0, and the two together reach 32.7 AP, 61.4 AP50, 29.6 AP75, and 46.1 AR on SODA-D, plus 35.8 AP and 44.7 AR on SODA-A. The paper concludes that balanced multi-clue assignment and category-aware feature interaction are complementary and jointly explain the gains.

Load-bearing premise

The balancing property rests on the claim in Section III-B3 that candidate-sample multi-clue confidence scores approximate a normal distribution; if they are skewed or multi-modal, the threshold $m_D+\gamma v_D$ will not keep the intended balance of positive samples across object sizes.

Editorial extensions

If this is right

  • Because MAFE R-CNN uses only the standard cross-entropy and smooth-$L_1$ losses, the two mechanisms can be dropped into other two-stage detectors without new loss terms or training schedules.
  • MCSS's size-dependent threshold implies that sample assignment should be calibrated per ground-truth size; fixed thresholds systematically under-supply tiny objects.
  • CFEM's category memory is fixed at inference, so the enhancement step adds only the cross-attention forward pass, with no ongoing memory-update cost.
  • The best configuration places the mechanisms at the second of three RoI heads, meaning the first stage should keep providing coarse candidates while the last stage refines, rather than applying the modules at every stage.
  • On the size-grouped SODA metrics the largest relative gains sit in the extremely-small band, matching the claim that balancing and category pooling help the smallest instances.

Reading between the lines

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

  • A testable extension is to attach MCSS and CFEM to assignment strategies and backbones other than Cascade R-CNN; the paper's ablations show the gains are complementary in this one architecture but do not establish transfer.
  • The category memory stores a single prototype per class, so it cannot represent intra-class appearance variation; multi-prototype or scale-conditioned memories are a natural next step that the paper leaves unexamined.
  • Because the balancing argument depends on the normality of confidence scores, a fully adaptive assigner that estimates its threshold from the empirical distribution would be a more robust variant if small-object scores turn out to be skewed.
  • The paper does not report inference latency; if the cross-attention enhancement is applied to all candidate boxes in a stage, real-time systems would need to restrict it to a shallow stage or top-ranked candidates.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 6 minor

Summary. The paper proposes MAFE R-CNN, a two-stage small object detector combining Multi-Clue Sample Selection (MCSS) and a Category-aware Feature Enhancement Mechanism (CFEM). MCSS selects positive samples using a weighted combination of classification confidence and IoU, with a dynamic threshold that depends on the ground-truth size and the mean/standard deviation of the multi-clue confidence. CFEM maintains a category-aware memory, updated from ground-truth features during training, and enhances candidate box features through weighted memory aggregation and cross-attention. Experiments on SODA-D and SODA-A report state-of-the-art AP and AR, with ablations attributing gains to both components.

Significance. If the reported results hold, the paper provides a practical, conceptually simple contribution to small object detection, with strong component-wise ablations (Tables III, VI–IX) and consistent gains on two challenging benchmarks. The combination of multi-clue sample assignment and cross-image category memory is novel relative to prior single-image feature enhancement approaches. However, the theoretical justification of the MCSS dynamic threshold rests on an unverified normality assumption, and the absence of variance reporting limits the certainty of the headline gains. The paper is likely to be of interest to the small-object detection community, but the main mechanism needs stronger empirical support.

major comments (2)
  1. [§III-B3, Eq. (4)] The claim that 'the confidence of candidate samples approximates a normal distribution' is central to the MCSS design, but it is not supported by any evidence in the paper. The multi-clue confidence D in Eq. (2) is a convex combination of a sigmoid classifier output and an IoU value, both bounded in [0,1], and for small objects the IoU component is typically low and right-skewed; with only k=9 candidates, the mean mD and standard deviation vD are also noisy. If the distribution is not approximately normal, the statement that mD + γ·vD yields roughly 0.16k positive samples (and the corresponding size-dependent balancing) does not follow. Figure 1(a) reports downstream assignment counts, but it does not directly verify the distributional assumption. Please provide empirical histograms or CDFs of D for different object-size groups and report how often the β cap in Eq. (4) is binding, or revise the mechanism description to match the actual empirical behavior.
  2. [Tables I and II] The headline results are reported as single runs without error bars or significance testing. The AP improvement over the second-best method (e.g., 1.5 AP over Cascade R-CNN on SODA-D) might be within run-to-run variability, especially given the large and diverse SODA dataset. Please report results over at least three random seeds with mean and standard deviation, or an equivalent significance measure, for the main comparison and for the core ablation in Table III.
minor comments (6)
  1. [§IV-A, Eq. (5)] The momentum m used in the category-aware memory update Eq. (5) is not specified in Implementation Details; please report its value.
  2. [Figure 2] The figure contains typos: 'MPSS' should be 'MCSS', and 'Feature Engancement' should be 'Feature Enhancement'.
  3. [§III-B1] The word 'numbner' should be 'number'.
  4. [References] The reference list appears twice at the end of the manuscript; please remove the duplicate.
  5. [Algorithm 1, line 12] The condition that the sample center must be inside the ground-truth box is introduced without discussion in the text; please clarify its motivation and effect.
  6. [Figure 1(a)] The caption does not explain the meaning of 'Avg Assign Num' or the units of 'Absolute Size'; consider adding a brief description.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: MAFE R-CNN is an empirical detector evaluated on external benchmarks, and its components are defined independently of the reported results.

full rationale

The paper's central claim is benchmark performance on SODA-D and SODA-A (Tables I-II). The two components, MCSS and CFEM, are defined by explicit equations (Eqs. 1-10) and trained with the standard detection loss (Eq. 11). The dynamic threshold in Eq. (4) is a designed assignment rule, not a fitted parameter disguised as a prediction; it uses the mean and standard deviation of the multi-clue confidence D, which is itself defined in Eq. (2) from IoU and classifier confidence. The claim that D is approximately normal is imported from ATSS [50] and is an unverified empirical assumption, which is a correctness/robustness risk rather than a circular step, because no reported number is derived from the normality assumption by construction. The ablation hyperparameters k, alpha, beta, and gamma are tuned on SODA-D, the same benchmark as the headline comparison, but the paper reports detector accuracy on the benchmark in the standard way; this is overfitting risk, not circularity. Self-citations ([7], [8], [34], [49]) are to the authors' prior work on infrared small-target detection, gigapixel tracking, multi-object tracking, and image completion, and none is load-bearing for the MCSS or CFEM derivation. No uniqueness theorem or prior result by the authors is invoked to force the design choice. Overall, the derivation chain is self-contained against external benchmarks, and no equation-level reduction of a prediction to its inputs is present.

Assumptions & free parameters 7 free parameters · 4 assumptions · 1 invented entities

The central claim rests on standard supervised learning assumptions and a small set of tuned hyperparameters. The most important free parameters are the MCSS hyperparameters (k, α, β, γ_max), which are selected via ablations on the same benchmark used for the final results. The category-aware memory is a new architectural component but does not introduce a new physical entity.

free parameters (7)
  • k = 9
    Number of candidate samples selected by center distance. Tuned via ablation on SODA-D (Table VI).
  • alpha = 0.3
    Weight for category confidence in multi-clue confidence. Tuned via ablation on SODA-D (Table VII).
  • beta = 0.6
    Minimum positive sample threshold. Chosen by hand/experimental observation, reported in Section IV-A.
  • gamma_max = 3.0
    Cap on standard ratio. Based on experimental observations, reported in Section IV-A.
  • Smax = 32
    Maximum absolute size of small objects, taken from prior literature (COCO/SODA), but still a constant in the threshold computation.
  • memory momentum m = not reported
    Momentum for the category-aware memory update (Eq. 5) is not specified in the text, leaving a hidden free parameter.
  • memory dimension D = 1024
    Dimension of the category-aware memory, set by hand in Section IV-A.
assumptions (4)
  • domain assumption Multi-clue confidence scores approximate a normal distribution
    Section III-B3 states 'The confidence of candidate samples approximates a normal distribution', and the threshold formula T_g = min(m_D + γ·v_D, β) depends on this to yield balanced sample counts.
  • domain assumption Larger objects have more candidate samples, so threshold should scale with object size
    Section III-B3 uses the ratio γ = min(S_g/S_max, 3) to adjust the threshold, assuming a linear relationship between region size and the needed threshold offset.
  • domain assumption Ground-truth features of the same category share structure that can be aggregated via cosine-similarity weighting
    Section III-C1 updates the memory using a cosine-similarity-weighted average of ground-truth features (Eq. 6-7), assuming this aggregation improves representativeness.
  • domain assumption The fixed category-aware memory at inference generalizes to test categories
    The memory is fixed after training and used for all inference (Section III-A and Figure 2), assuming the stored prototypes remain valid for unseen images.
invented entities (1)
  • Category-aware memory module M
    purpose: Store and aggregate category-level feature prototypes to enhance small object features via weighted combination and cross-attention.
    The module is a learned component whose effectiveness is shown only through in-paper ablations (Tables III and VIII). There is no external falsifiable prediction beyond the SODA benchmark results.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MAFE R-CNN: Selecting More Samples to Learn Category-aware Features for Small Object Detection." pith.science (2026). https://pith.science/paper/XVRKKRYV

@misc{pith2026250516442,
  author       = {Pith},
  title        = {Pith review of: MAFE R-CNN: Selecting More Samples to Learn Category-aware Features for Small Object Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XVRKKRYV}},
  note         = {Machine review of arXiv:2505.16442}
}
read the original abstract

Small object detection in intricate environments has consistently represented a major challenge in the field of object detection. In this paper, we identify that this difficulty stems from the detectors' inability to effectively learn discriminative features for objects of small size, compounded by the complexity of selecting high-quality small object samples during training, which motivates the proposal of the Multi-Clue Assignment and Feature Enhancement R-CNN.Specifically, MAFE R-CNN integrates two pivotal components.The first is the Multi-Clue Sample Selection (MCSS) strategy, in which the Intersection over Union (IoU) distance, predicted category confidence, and ground truth region sizes are leveraged as informative clues in the sample selection process. This methodology facilitates the selection of diverse positive samples and ensures a balanced distribution of object sizes during training, thereby promoting effective model learning.The second is the Category-aware Feature Enhancement Mechanism (CFEM), where we propose a simple yet effective category-aware memory module to explore the relationships among object features. Subsequently, we enhance the object feature representation by facilitating the interaction between category-aware features and candidate box features.Comprehensive experiments conducted on the large-scale small object dataset SODA validate the effectiveness of the proposed method. The code will be made publicly available.

Figures

Figures reproduced from arXiv: 2505.16442 by the authors.

Figure 1
Figure 1. (a) The average number of samples assigned by different assigners for [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the proposed MAFE R-CNN. Our method integrates Multi-Clue Sample Selection (MCSS) and Category-aware Feature Enhancement Mechanism (CFEM) into the MAFE R-CNN for training. During inference, MCSS is not used for sample selection, and the category-aware memory remains fixed. The three-stage prediction process based on the input features of small objects produces the final prediction results. B. Memory … view at source ↗
Figure 3
Figure 3. The illustration of CFEM training pipeline. Ground truth features [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Visualization on the SODA test set. Only small objects with a confidence score larger than 0.3 are shown. Best viewed in color and zoom-in windows. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Comparison of proposal feature distributions before (a) and after (b) [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

72 extracted references · 65 canonical work pages

  1. [1]

    Attentive alignment network for multispectral pedestrian detection,

    N. Chen, J. Xie, J. Nie, J. Cao, Z. Shao, and Y . Pang, “Attentive alignment network for multispectral pedestrian detection,” in ACM International Conference on Multimedia , 2023, pp. 3787–3795

    work page 2023

  2. [2]

    To- wards large-scale small object detection: Survey and benchmarks,

    G. Cheng, X. Yuan, X. Yao, K. Yan, Q. Zeng, X. Xie, and J. Han, “To- wards large-scale small object detection: Survey and benchmarks,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, no. 11, pp. 13 467–13 488, 2023

    2023

  3. [3]

    Scale match for tiny person detection,

    X. Yu, Y . Gong, N. Jiang, Q. Ye, and Z. Han, “Scale match for tiny person detection,” in IEEE Winter Conference on Applications of Computer Vision, 2020, pp. 1257–1265

    work page 2020

  4. [4]

    Rfla: Gaussian receptive field based label assignment for tiny object detection,

    C. Xu, J. Wang, W. Yang, H. Yu, L. Yu, and G.-S. Xia, “Rfla: Gaussian receptive field based label assignment for tiny object detection,” in European Conference on Computer Vision , 2022, pp. 526–543

    work page 2022

  5. [5]

    Visdrone-det2021: The vision meets drone object detection challenge results,

    Y . Cao, Z. He, L. Wang, W. Wang, Y . Yuan, D. Zhang, J. Zhang, P. Zhu, L. Van Gool, J. Han et al., “Visdrone-det2021: The vision meets drone object detection challenge results,” in IEEE International Conference on Computer Vision, 2021, pp. 2847–2854

    work page 2021

  6. [6]

    Accurate and robust object detection via selective adversarial learning with constraints,

    J. Chen, H. Li, Q. Gao, J. Liang, R. Zhang, L. Yin, and X. Chai, “Accurate and robust object detection via selective adversarial learning with constraints,” IEEE Transactions on Image Processing , pp. 5593– 5605, 2024. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 13

    work page 2024

  7. [7]

    Infrared small target detection with scale and location sensitivity,

    Q. Liu, R. Liu, B. Zheng, H. Wang, and Y . Fu, “Infrared small target detection with scale and location sensitivity,” in IEEE Conference on Computer Vision and Pattern Recognition , 2024, pp. 17 490–17 499

    work page 2024

  8. [8]

    Sliding window detection and distance-based matching for tracking on gigapixel images,

    Y . Li, Q. Liu, X. Wang, and Y . Fu, “Sliding window detection and distance-based matching for tracking on gigapixel images,” in CAAI International Conference on Artificial Intelligence , 2023

    work page 2023

Show all 72 references
  1. [9]

    Object detection in autonomous vehicles: Status and open challenges,

    A. Balasubramaniam and S. Pasricha, “Object detection in autonomous vehicles: Status and open challenges,” arXiv preprint arXiv:2201.07706, 2022

    2022 arXiv

  2. [10]

    Dot distance for tiny object detection in aerial images,

    C. Xu, J. Wang, W. Yang, and L. Yu, “Dot distance for tiny object detection in aerial images,” in IEEE Conference on Computer Vision and Pattern Recognition, 2021, pp. 1192–1201

    2021

  3. [11]

    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,” Advances in Neural Information Processing Systems , vol. 39, no. 6, 2015

    2015

  4. [12]

    Cascade r-cnn: High quality object detection and instance segmentation,

    Z. Cai and N. Vasconcelos, “Cascade r-cnn: High quality object detection and instance segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 5, pp. 1483–1498, 2019

    2019

  5. [13]

    Small object detection via coarse-to-fine proposal generation and imitation learning,

    X. Yuan, G. Cheng, K. Yan, Q. Zeng, and J. Han, “Small object detection via coarse-to-fine proposal generation and imitation learning,” in IEEE International Conference on Computer Vision , 2023, pp. 6317–6327

    2023

  6. [14]

    S3fd: Single shot scale-invariant face detector,

    S. Zhang, X. Zhu, Z. Lei, H. Shi, X. Wang, and S. Z. Li, “S3fd: Single shot scale-invariant face detector,” in IEEE International Conference on Computer Vision, 2017, pp. 192–201

    2017

  7. [15]

    Seeing small faces from robust anchor’s perspective,

    C. Zhu, R. Tao, K. Luu, and M. Savvides, “Seeing small faces from robust anchor’s perspective,” in IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 5127–5136

    2018

  8. [16]

    Scale-aware fast r-cnn for pedestrian detection,

    J. Li, X. Liang, S. Shen, T. Xu, J. Feng, and S. Yan, “Scale-aware fast r-cnn for pedestrian detection,” IEEE Transactions on Multimedia , vol. 20, no. 4, pp. 985–996, 2017

    2017

  9. [17]

    Feature pyramid networks for object detection,

    T.-Y . Lin, P. Doll´ar, R. Girshick, K. He, B. Hariharan, and S. Belongie, “Feature pyramid networks for object detection,” in IEEE Conference on Computer Vision and Pattern Recognition , 2017, pp. 2117–2125

    2017

  10. [18]

    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 Conference on Computer Vision and Pattern Recognition , 2022, pp. 13 668–13 677

    2022

  11. [19]

    Finding tiny faces in the wild with generative adversarial network,

    Y . Bai, Y . Zhang, M. Ding, and B. Ghanem, “Finding tiny faces in the wild with generative adversarial network,” in IEEE Conference on Computer Vision and Pattern Recognition , 2018, pp. 21–30

    2018

  12. [20]

    Sod-mtgan: Small object detection via multi-task generative adversarial network,

    ——, “Sod-mtgan: Small object detection via multi-task generative adversarial network,” in European Conference on Computer Vision , 2018, pp. 206–221

    2018

  13. [21]

    Better to follow, follow to be better: Towards precise supervision of feature super-resolution for small object detection,

    J. Noh, W. Bae, W. Lee, J. Seo, and G. Kim, “Better to follow, follow to be better: Towards precise supervision of feature super-resolution for small object detection,” in IEEE International Conference on Computer Vision, 2019, pp. 9725–9734

    2019

  14. [22]

    Extended feature pyramid network for small object detection,

    C. Deng, M. Wang, L. Liu, Y . Liu, and Y . Jiang, “Extended feature pyramid network for small object detection,” IEEE Transactions on Multimedia, vol. 24, pp. 1968–1979, 2021

    1968

  15. [23]

    Path aggregation network for instance segmentation,

    S. Liu, L. Qi, H. Qin, J. Shi, and J. Jia, “Path aggregation network for instance segmentation,” in IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 8759–8768

    2018

  16. [24]

    M2det: A single-shot object detector based on multi-level feature pyramid network,

    Q. Zhao, T. Sheng, Y . Wang, Z. Tang, Y . Chen, L. Cai, and H. Ling, “M2det: A single-shot object detector based on multi-level feature pyramid network,” in AAAI Conference on Artificial Intelligence, vol. 33, no. 01, 2019, pp. 9259–9266

    2019

  17. [25]

    Effective fusion factor in fpn for tiny object detection,

    Y . Gong, X. Yu, Y . Ding, X. Peng, J. Zhao, and Z. Han, “Effective fusion factor in fpn for tiny object detection,” in IEEE Winter Conference on Applications of Computer Vision , 2021, pp. 1160–1168

    2021

  18. [26]

    Self-mimic learning for small-scale pedestrian detection,

    J. Wu, C. Zhou, Q. Zhang, M. Yang, and J. Yuan, “Self-mimic learning for small-scale pedestrian detection,” in ACM International Conference on Multimedia, 2020, pp. 2012–2020

    2020

  19. [27]

    Robust small-scale pedestrian detection with cued recall via memory learning,

    J. U. Kim, S. Park, and Y . M. Ro, “Robust small-scale pedestrian detection with cued recall via memory learning,” in IEEE International Conference on Computer Vision , 2021, pp. 3050–3059

    2021

  20. [28]

    Mlfa: Towards realistic test time adaptive object detection by multi-level feature alignment,

    Y . Liu, J. Wang, C. Huang, Y . Wu, Y . Xu, and X. Cao, “Mlfa: Towards realistic test time adaptive object detection by multi-level feature alignment,” IEEE Transactions on Image Processing , pp. 5837– 5848, 2024

    2024

  21. [29]

    Rich feature hierarchies for accurate object detection and semantic segmentation,

    R. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,” in IEEE Conference on Computer Vision and Pattern Recognition , 2014, pp. 580–587

    2014

  22. [30]

    Centernet: Keypoint triplets for object detection,

    K. Duan, S. Bai, L. Xie, H. Qi, Q. Huang, and Q. Tian, “Centernet: Keypoint triplets for object detection,” in IEEE International Conference on Computer Vision , 2019, pp. 6569–6578

    2019

  23. [31]

    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,”arXiv preprint arXiv:2010.04159, 2020

    2010 arXiv

  24. [32]

    Yolox: Exceeding yolo series in 2021,

    Z. Ge, S. Liu, F. Wang, Z. Li, and J. Sun, “Yolox: Exceeding yolo series in 2021,” arXiv preprint arXiv:2107.08430 , 2021

    2021 arXiv

  25. [33]

    Safnet: A semi-anchor-free network with enhanced feature pyramid for object detection,

    Z. Jin, B. Liu, Q. Chu, and N. Yu, “Safnet: A semi-anchor-free network with enhanced feature pyramid for object detection,” IEEE Transactions on Image Processing , vol. 29, pp. 9445–9457, 2020

    2020

  26. [34]

    Siamese-detr for generic multi- object tracking,

    Q. Liu, Y . Li, Y . Jiang, and Y . Fu, “Siamese-detr for generic multi- object tracking,” IEEE Transactions on Image Processing , vol. 33, pp. 3935–3949, 2024

    2024

  27. [35]

    Frequency- aware feature fusion for dense image prediction,

    L. Chen, Y . Fu, L. Gu, C. Yan, T. Harada, and G. Huang, “Frequency- aware feature fusion for dense image prediction,” IEEE Transactions on Pattern Analysis and Machine Intelligence , pp. 1–18, 2024

    2024

  28. [36]

    Stairnet: Top-down semantic aggregation for accurate one shot detection,

    S. Woo, S. Hwang, and I. S. Kweon, “Stairnet: Top-down semantic aggregation for accurate one shot detection,” in IEEE Winter Conference on Applications of Computer Vision , 2018, pp. 1093–1102

    2018

  29. [37]

    Ipg-net: Image pyramid guidance network for small object detection,

    Z. Liu, G. Gao, L. Sun, and L. Fang, “Ipg-net: Image pyramid guidance network for small object detection,” in IEEE Conference on Computer Vision and Pattern Recognition workshops , 2020, pp. 1026–1027

    2020

  30. [38]

    Centralized feature pyramid for object detection,

    Y . Quan, D. Zhang, L. Zhang, and J. Tang, “Centralized feature pyramid for object detection,” IEEE Transactions on Image Processing , vol. 32, pp. 4341–4354, 2023

    2023

  31. [39]

    Perceptual generative adversarial networks for small object detection,

    J. Li, X. Liang, Y . Wei, T. Xu, J. Feng, and S. Yan, “Perceptual generative adversarial networks for small object detection,” in IEEE Conference on Computer Vision and Pattern Recognition , 2017, pp. 1222–1230

    2017

  32. [40]

    Mega: molecular evolutionary genetics analysis software for microcomputers,

    S. Kumar, K. Tamura, and M. Nei, “Mega: molecular evolutionary genetics analysis software for microcomputers,” Bioinformatics, vol. 10, no. 2, pp. 189–191, 1994

    1994

  33. [41]

    Mcibi++: Soft mining contextual information beyond image for semantic segmentation,

    Z. Jin, D. Yu, Z. Yuan, and L. Yu, “Mcibi++: Soft mining contextual information beyond image for semantic segmentation,” IEEE Transac- tions on Pattern Analysis and Machine Intelligence , vol. 45, no. 5, pp. 5988–6005, 2022

    2022

  34. [42]

    Mne software for processing meg and eeg data,

    A. Gramfort, M. Luessi, E. Larson, D. A. Engemann, D. Strohmeier, C. Brodbeck, L. Parkkonen, and M. S. H ¨am¨al¨ainen, “Mne software for processing meg and eeg data,” neuroimage, vol. 86, pp. 446–460, 2014

    2014

  35. [43]

    Cross-batch memory for embedding learning,

    X. Wang, H. Zhang, W. Huang, and M. R. Scott, “Cross-batch memory for embedding learning,” in IEEE Conference on Computer Vision and Pattern Recognition, 2020, pp. 6388–6397

    2020

  36. [44]

    Memory enhanced global- local aggregation for video object detection,

    Y . Chen, Y . Cao, H. Hu, and L. Wang, “Memory enhanced global- local aggregation for video object detection,” in IEEE Conference on Computer Vision and Pattern Recognition , 2020, pp. 10 337–10 346

    2020

  37. [45]

    Invariance matters: Exemplar memory for domain adaptive person re-identification,

    Z. Zhong, L. Zheng, Z. Luo, S. Li, and Y . Yang, “Invariance matters: Exemplar memory for domain adaptive person re-identification,” inIEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 598– 607

    2019

  38. [46]

    Memory-based neigh- bourhood embedding for visual recognition,

    S. Li, D. Chen, B. Liu, N. Yu, and R. Zhao, “Memory-based neigh- bourhood embedding for visual recognition,” in IEEE International Conference on Computer Vision , 2019, pp. 6102–6111

    2019

  39. [47]

    Long-term feature banks for detailed video understanding,

    C.-Y . Wu, C. Feichtenhofer, H. Fan, K. He, P. Krahenbuhl, and R. Gir- shick, “Long-term feature banks for detailed video understanding,” in IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 284–293

    2019

  40. [48]

    Object detection difficulty: Suppressing over-aggregation for faster and better video object detection,

    B. Zhang, S. Wang, Y . Liu, B. Kusy, X. Li, and J. Liu, “Object detection difficulty: Suppressing over-aggregation for faster and better video object detection,” in ACM International Conference on Multimedia , 2023, pp. 1768–1778

    2023

  41. [49]

    Transformer based pluralistic image completion with reduced information loss,

    Q. Liu, Y . Jiang, Z. Tan, D. Chen, Y . Fu, Q. Chu, G. Hua, and N. Yu, “Transformer based pluralistic image completion with reduced information loss,” IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 6652–6668, 2024

    2024

  42. [50]

    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 Conference on Computer Vision and Pattern Recognition, 2020, pp. 9759–9768

    2020

  43. [51]

    Fast r-cnn,

    R. Girshick, “Fast r-cnn,” in IEEE International Conference on Com- puter Vision, 2015, pp. 1440–1448

    2015

  44. [52]

    Crossvit: Cross-attention multi- scale vision transformer for image classification,

    C.-F. R. Chen, Q. Fan, and R. Panda, “Crossvit: Cross-attention multi- scale vision transformer for image classification,” in IEEE International Conference on Computer Vision , 2021, pp. 357–366

    2021

  45. [53]

    Cornernet: Detecting objects as paired keypoints,

    H. Law and J. Deng, “Cornernet: Detecting objects as paired keypoints,” in European Conference on Computer Vision , 2018, pp. 734–750

    2018

  46. [54]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 770–778. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    2016

  47. [55]

    Reppoints: Point set representation for object detection,

    Z. Yang, S. Liu, H. Hu, L. Wang, and S. Lin, “Reppoints: Point set representation for object detection,” in IEEE International Conference on Computer Vision , 2019, pp. 9657–9666

    2019

  48. [56]

    Sparse r-cnn: End-to-end object detection with learnable proposals,

    P. Sun, R. Zhang, Y . Jiang, T. Kong, C. Xu, W. Zhan, M. Tomizuka, L. Li, Z. Yuan, C. Wang et al. , “Sparse r-cnn: End-to-end object detection with learnable proposals,” in IEEE Conference on Computer Vision and Pattern Recognition , 2021, pp. 14 454–14 463

    2021

  49. [57]

    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 Conference on Computer Vision and Pattern Recognition , 2024

    2024

  50. [58]

    Focal loss for dense object detection,

    T.-Y . Lin, P. Goyal, R. Girshick, K. He, and P. Doll ´ar, “Focal loss for dense object detection,” in IEEE International Conference on Computer Vision, 2017, pp. 2980–2988

    2017

  51. [59]

    Fcos: Fully convolutional one- stage object detection,

    Z. Tian, C. Shen, H. Chen, and T. He, “Fcos: Fully convolutional one- stage object detection,” in IEEE International Conference on Computer Vision, 2019, pp. 9627–9636

    2019

  52. [60]

    Dynamic head: Unifying object detection heads with attentions,

    X. Dai, Y . Chen, B. Xiao, D. Chen, M. Liu, L. Yuan, and L. Zhang, “Dynamic head: Unifying object detection heads with attentions,” in IEEE Conference on Computer Vision and Pattern Recognition , 2021, pp. 7373–7382

    2021

  53. [61]

    Cascade rpn: Delving into high-quality region proposal network with adaptive convolution,

    T. Vu, H. Jang, T. X. Pham, and C. Yoo, “Cascade rpn: Delving into high-quality region proposal network with adaptive convolution,” Advances in Neural Information Processing Systems , vol. 32, 2019

    2019

  54. [62]

    Kldet: Detecting tiny objects in remote sensing im- ages via kullback-leibler divergence,

    Z. Zhou and Y . Zhu, “Kldet: Detecting tiny objects in remote sensing im- ages via kullback-leibler divergence,” IEEE Transactions on Geoscience and Remote Sensing , 2024

    2024

  55. [63]

    Align deep features for oriented object detection,

    J. Han, J. Ding, J. Li, and G.-S. Xia, “Align deep features for oriented object detection,” IEEE Transactions on Geoscience and Remote Sens- ing, vol. 60, pp. 1–11, 2021

    2021

  56. [64]

    Oriented reppoints for aerial object detection,

    W. Li, Y . Chen, K. Hu, and J. Zhu, “Oriented reppoints for aerial object detection,” in IEEE Conference on Computer Vision and Pattern Recognition, 2022, pp. 1829–1838

    2022

  57. [65]

    Multi-oriented object detection in aerial im- ages with double horizontal rectangles,

    G. Nie and H. Huang, “Multi-oriented object detection in aerial im- ages with double horizontal rectangles,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 4, pp. 4932–4944, 2022

    2022

  58. [66]

    Gliding vertex on the horizontal bounding box for multi-oriented object detection,

    Y . Xu, M. Fu, Q. Wang, Y . Wang, K. Chen, G.-S. Xia, and X. Bai, “Gliding vertex on the horizontal bounding box for multi-oriented object detection,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 4, pp. 1452–1459, 2020

    2020

  59. [67]

    Oriented r-cnn and beyond,

    X. Xie, G. Cheng, J. Wang, K. Li, X. Yao, and J. Han, “Oriented r-cnn and beyond,” International Journal of Computer Vision, pp. 1–23, 2024

    2024

  60. [68]

    Dual-aligned oriented detector,

    G. Cheng, Y . Yao, S. Li, K. Li, X. Xie, J. Wang, X. Yao, and J. Han, “Dual-aligned oriented detector,” IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–11, 2022

    2022

  61. [69]

    Lsknet: A foundation lightweight backbone for remote sensing,

    Y . Li, X. Li, Y . Dai, Q. Hou, L. Liu, Y . Liu, M.-M. Cheng, and J. Yang, “Lsknet: A foundation lightweight backbone for remote sensing,” arXiv preprint arXiv:2403.11735, 2024

    2024 arXiv

  62. [70]

    Dota: A large-scale dataset for object detection in aerial images,

    G.-S. Xia, X. Bai, J. Ding, Z. Zhu, S. Belongie, J. Luo, M. Datcu, M. Pelillo, and L. Zhang, “Dota: A large-scale dataset for object detection in aerial images,” in IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 3974–3983

    2018

  63. [71]

    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 European Conference on Computer Vision , 2014, pp. 740– 755

    2014

  64. [72]

    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 European Conference on Computer Vision , 2020, pp. 213–229

    2020

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.