Pith. sign in

REVIEW 3 major objections 5 minor 38 references

RemDet: Rethinking Efficient Model Design for UAV Object Detection

T0 review · 3 major / 5 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 RemDet, a one-stage anchor-free detector built around the principle of reducing information loss, claims state-of-the-art accuracy on VisDrone while running at 110 FPS on a single GPU.

desk verdict The base detector is a credible incremental contribution, but the headline 40 mAP @ 9 ms o+ca row conflates single-image latency with the cluster-crop pipeline and needs clarification before the SOTA claim is taken at face value. read the letter →

arxiv 2412.10040 v2 pith:55O3GYCF submitted 2024-12-13 cs.CV

Chen Li , Rui Zhao , Zeyu Wang , Huiying Xu , Xinzhong Zhu This is my paper · ORCID
classification cs.CV
keywords UAVobjectdetectionsmallreal-timeinformationbottleneckgatedfeed-forwardnetworkreparameterizationVisDronedownsampling
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

RemDet is a one-stage anchor-free detector that aims to show that real-time UAV object detection can be improved by treating information loss as the main design criterion, rather than by adding handcrafted coarse-to-fine pipelines. It claims that widening channels and using multiplication-based gating give high-dimensional representations that preserve small-object information at low latency. On VisDrone, the largest model reaches 40.0 mAP at 9 ms per image, improving on the previous best result by more than 3.4 points while running at 110 FPS.

What carries the argument

The load-bearing object is GatedFFN, a block that computes two pointwise projections of the input, applies GELU, multiplies them element-wise, then fuses with a 1x1 convolution and reparameterized depthwise convolutions. The paper's analytic point is that multiplication of two projections produces a polynomial with O($d^{2}$) distinct terms at no extra parameter cost compared with summing the same projections, so gating gives a higher-dimensional representation than a standard feed-forward network. The C2f-style ChannelC2f and CED downsample modules are support structures that extend the same information-preservation principle to neck width and resolution reduction.

What would settle it

Run RemDet-X at 1024 on the original VisDrone validation set without cluster-aware crops, or run UFPMP-Det with the same crop settings; if the 3.4-point margin disappears, the headline state-of-the-art claim does not hold. In the same setup, compare GatedFFN against ConvFFN at a fixed accuracy budget; if the multiplication advantage reverses at expansion factors above those tested, the cost-effectiveness claim is empirically bounded.

Watch

Extended reading notes

Core claim

The paper's central claim is that information loss, not model capacity or handcrafted region proposal, is the main bottleneck for small, dense object detection in UAV images; consequently, architectures should be designed to preserve information across layers. Three modules implement this: ChannelC2f widens the neck channels, CED replaces simple stride-2 convolutions with a combined depthwise convolution and patch-merge downsampling, and GatedFFN replaces the standard feed-forward block with a reparameterized gated multiplication branch. The paper argues analytically and empirically that element-wise multiplication of two projected feature vectors yields an implicit polynomial expansion with more distinct terms than a comparable MLP, making it a cheaper way to reach high-dimensional representations. On VisDrone, RemDet-Tiny improves over the YOLOv8 baseline by 2.7 mAP at comparable speed, and RemDet-X with cluster-aware crops sets 40.0 mAP at 9 ms latency.

Load-bearing premise

The reported state-of-the-art comparison assumes that using cluster-aware crops for RemDet-X while most baselines are evaluated on the original validation set is a fair protocol, and that tuning stage ratios and expansion factors on the VisDrone validation set does not inflate the reported mAP.

Editorial extensions

If this is right

  • RemDet-Tiny can serve as a drop-in real-time detector for embedded UAV use, beating YOLO-family detectors of similar size on VisDrone.
  • The information-loss guideline transfers to other detection backbones: replacing backbones in Faster R-CNN, RetinaNet, and DyHead improves their VisDrone AP.
  • GatedFFN's multiplication principle suggests that gated multiplicative interactions are a lower-latency route to high-dimensional feature representation than wider MLPs.
  • The combination with cluster-aware crops shows that handcrafted coarse-to-fine pipelines can be replaced or reduced by a simpler architecture that keeps high speed.

Reading between the lines

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

  • The information-loss explanation is underdetermined by the experiments; widening channels and multiplication also change parameter count and training dynamics, so the causal role of 'information loss' specifically would need controlled trials.
  • The paper's own caveat that large-object AP lags behind suggests the approach is tuned to small-object distributions; on datasets with large objects the gains may shrink or reverse.
  • A natural test is to ablate GatedFFN against a ConvFFN with equal FLOPs and latency across expansion factors 3, 6, and 9 on a non-UAV dataset to see whether the multiplication advantage is dataset-specific.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. RemDet proposes a real-time UAV object detector built around three components: GatedFFN (a gated multiplication block that reparameterizes two convolutions), ChannelC2f (a wider C2f variant for the neck), and Context Enhanced Downsample (CED, a ViT-inspired patch-merge downsampling module). The paper motivates these designs through an 'information loss' principle derived from information bottleneck theory, and claims state-of-the-art results on VisDrone with a 3.4% improvement over prior methods while reaching 110 FPS on a single RTX 4090. Experiments are reported on VisDrone, UAVDT, and MSCOCO, together with ablations and combinations of the RemDet backbone with other detectors.

Significance. If the reported results hold, the paper makes a useful practical contribution: it shows that a relatively simple, hardware-friendly architecture can match or exceed more complex coarse-to-fine UAV detectors, and the cross-dataset experiments on MSCOCO and UAVDT provide evidence of generalization. The paper also ships code and fairly extensive ablations, which is a strength. However, the headline claim of simultaneous state-of-the-art accuracy and real-time speed rests on an evaluation protocol whose overhead is not described, and the theoretical 'information loss' argument is not rigorous. These issues are fixable but currently make the central claim partly unsubstantiated.

major comments (3)
  1. [Table 1 and 'Comparison Results on VisDrone'] The headline configuration RemDet-X 1024 o+ca reports 9.0 ms latency and 182 GFLOPs, but the paper never specifies whether these figures include the cluster-aware crop (ca) pipeline. In prior o+ca methods (YOLC, UFPMP-Det) the crop generation and multiple network passes are part of the reported latency (441 ms and 152 ms respectively). If the 9.0 ms is single-image latency, then the 40.0 mAP result is not real-time in the same end-to-end sense, and the 110 FPS claim does not apply to the SOTA configuration; if it is end-to-end, the ca procedure must be radically cheaper than in prior work, yet no description, pseudocode, or ablation of the crop pipeline is given. Either way the central claim that one configuration achieves both SOTA accuracy and real-time speed is not supported by the reported numbers.
  2. [Section 3.2, Eq. (3)-(4)] The polynomial argument for multiplication is not a rigorous proof that multiplication gives 'higher dimensionality' or better representations. Counting (d+1)d/2 distinct monomials treats the network as computing symbolic polynomials, but in a learned linear layer the weights are not independent symbols, and the output of Eq. (4) is still a single scalar per output channel. The claim that wT2x is 'discarded but implicitly included' is not formalized, and the conclusion that multiplication is 'more cost-effective than feed-forward networks' is not supported by any latency or FLOP comparison of the two modules at matched accuracy; Figure 4(b) reports only mAP versus expansion, not cost-effectiveness. The empirical results may stand, but the theoretical story needs either a precise statement or a re-framing as intuition.
  3. [Ablation of Overall Design; Figure 7; Appendix F] The final architecture choices, including the stage block ratio (3:3:6:3), the GatedFFN expansion factor 3, and the CED channel expansion ratio, are selected by ablations on the VisDrone validation set, and the SOTA numbers in Table 1 are reported on the same validation set. This makes the 3.4% improvement over prior work vulnerable to validation-set overfitting; there is no held-out test set or repeated evaluation protocol. The COCO and UAVDT results provide some cross-dataset evidence, but they do not resolve the VisDrone-specific tuning issue. Please either report results on the VisDrone test set (if available), use a validation split distinct from the tuning split, or at least quantify the sensitivity of the headline gain to the tuned hyperparameters.
minor comments (5)
  1. [Throughout] The paper consistently types 'UA V' instead of 'UAV' (Abstract, Section 1, Figure 1 caption); this should be corrected.
  2. [Figure 3] The label 'Marco Design' appears to be a typo for 'Macro Design'; please fix.
  3. [Table 1] The footnote defines o, ca, and aug but does not explain the bare 'ca' entry for CDMNet; please clarify whether 'ca' implies 'cluster-aware' evaluation without the original validation set, and how that differs from 'o+ca'.
  4. [Appendix B] The statement that Multiplication 'discards the wT2x part' seems to contradict the main-text claim that wT2x is implicitly included in the polynomial output; please reconcile this wording.
  5. [Table 2] RemDet-L is marked with a dagger for cluster-aware crops, but no latency or FLOPs are given for that configuration; since the UAVDT comparison table reports only AP, please state whether efficiency was measured and, if not, note its absence explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: RemDet's central claims are empirical comparisons on external benchmarks, and the design rationale is not defined in terms of the reported results.

full rationale

The paper's central claims are empirical: RemDet is compared on VisDrone, UAVDT, and MSCOCO against external baselines, so the reported accuracy is not derived from its own assumptions by construction. The 'information loss' guideline is an adaptation of the external information bottleneck theory (Tishby and Zaslavsky; Shwartz-Ziv and Tishby) and is used only as a heuristic motivation for module designs, not as a fitted parameter that later reappears as a 'prediction'. The GatedFFN analysis in Section 3.2 is a self-contained algebraic comparison of MLP versus multiplication; its conclusion is independently supported by the ablation experiments in Figure 4(b) and Table 6. No load-bearing self-citations appear in the derivation chain, and no uniqueness theorem or prior same-author result is invoked to force the architecture choice. The main validity concerns are experimental rather than circular: some hyperparameters (stage ratios and channel expansions) were selected using the VisDrone validation set, and the reported 9 ms latency for the o+ca configuration is not fully specified. These issues affect fairness and reproducibility of the empirical claims, but they do not make the derivation equivalent to its inputs. Therefore, no significant circularity is present.

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

The paper introduces no new physical or conceptual entities; 'GatedFFN', 'ChannelC2f', and 'CED' are architectural modules, not postulates with independent evidence requirements. The load-bearing free parameters are architecture hyperparameters tuned on the evaluation dataset, and the theoretical framework is used as post hoc motivation rather than a predictive derivation.

free parameters (4)
  • Channel expansion factor for GatedFFN (e = 3) = 3
    Chosen based on experiments in Figure 4(b); the paper states 'we set the channel expansion factor to 3' after comparing expansions 1-9.
  • Stage block ratios = 3:3:6:3
    Initial ratios (3:6:6:3) from NAS on MSCOCO, then adjusted by ablations on VisDrone to remove blocks from stage 2 (Appendix F).
  • ChannelC2f expansion and bottleneck ratio = expansion 1.0, bottleneck ratio 0.25
    Set by hand to enhance intra-layer information while minimizing dense computation, Section 3.3.
  • CED channel expansion ratio = 1 (stage 1 expanded to 2)
    The paper reports that increasing from 1 to 2 improves performance but increases cost; they chose a balance and only apply expansion to stage 1 (Figure 7).
assumptions (4)
  • domain assumption The information bottleneck Lagrangian and the data processing inequality chain (Eqs. 1-2) justify minimizing information loss as the design guideline.
    These standard information-theoretic statements are used as motivation, but the paper never measures the mutual information terms in the trained detector; it assumes the principle transfers to architecture design.
  • ad hoc to paper Higher channel dimensionality reduces information loss and therefore improves small-object detection.
    This is the paper's central design principle, supported only by its own feature visualizations and ablations; it is not independently established.
  • ad hoc to paper Element-wise multiplication (Eq. 4) yields a more cost-effective high-dimensional representation than MLP addition.
    The polynomial term count is correct, but the claim that this translates to better accuracy at equal cost is supported only by the authors' experiments on their own architecture.
  • domain assumption Latency on a 4090 is a representative measure of real-time deployment performance.
    The paper reports FPS on one high-end GPU without specifying batch size, precision, or framework optimization, yet uses it as a headline efficiency claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RemDet: Rethinking Efficient Model Design for UAV Object Detection." pith.science (2026). https://pith.science/paper/55O3GYCF

@misc{pith2026241210040,
  author       = {Pith},
  title        = {Pith review of: RemDet: Rethinking Efficient Model Design for UAV Object Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/55O3GYCF}},
  note         = {Machine review of arXiv:2412.10040}
}
read the original abstract

Object detection in Unmanned Aerial Vehicle (UAV) images has emerged as a focal area of research, which presents two significant challenges: i) objects are typically small and dense within vast images; ii) computational resource constraints render most models unsuitable for real-time deployment. Current real-time object detectors are not optimized for UAV images, and complex methods designed for small object detection often lack real-time capabilities. To address these challenges, we propose a novel detector, RemDet (Reparameter efficient multiplication Detector). Our contributions are as follows: 1) Rethinking the challenges of existing detectors for small and dense UAV images, and proposing information loss as a design guideline for efficient models. 2) We introduce the ChannelC2f module to enhance small object detection performance, demonstrating that high-dimensional representations can effectively mitigate information loss. 3) We design the GatedFFN module to provide not only strong performance but also low latency, effectively addressing the challenges of real-time detection. Our research reveals that GatedFFN, through the use of multiplication, is more cost-effective than feed-forward networks for high-dimensional representation. 4) We propose the CED module, which combines the advantages of ViT and CNN downsampling to effectively reduce information loss. It specifically enhances context information for small and dense objects. Extensive experiments on large UAV datasets, Visdrone and UAVDT, validate the real-time efficiency and superior performance of our methods. On the challenging UAV dataset VisDrone, our methods not only provided state-of-the-art results, improving detection by more than 3.4%, but also achieve 110 FPS on a single 4090.

Figures

Figures reproduced from arXiv: 2412.10040 by the authors.

Figure 1
Figure 1. (a) Visualization of real-time detector results on [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed RemDet. Our method consists of three components: GatedFFN achieves adaptive weighting [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. RemDet design process evaluated using FLOP, [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: (a) Visualization of model features as channel expansion increases. (b) Comparison of Multiplication and ConvFFN. [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Structure of the IB, C2f, and proposed GatedFFN. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Visualization of the results of RetinaNet, CEASC, GFL and RemDet on the VisDrone and UAVDT datasets. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Ablation for each stage using different expansions. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Compare Multiplication and with w T 2 x [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Ablation of Backbone’s stage ratios. C. Ablation Experiments of RemDet When conducting ablation studies on the VisDrone dataset, we analyzed the impact of each component on detection accu￾racy and computational complexity, as shown in [PITH_FULL_IMAGE:figures/full_fig…
Figure 10
Figure 10. Figure 10: The detailed designs of CED module. more dense computation. Our method effectively mitigates this dense computation, resulting in a wider overall model. Importantly, the results demonstrate that at larger scales, our approach not only maintains a precision advantage b…
Figure 11
Figure 11. Figure 11: More visualized results of RetinaNet, CEASC, GFL and RemDet on the VisDrone. [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 21 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    W.; Sultani, W.; and Shah, M

    Ashraf, M. W.; Sultani, W.; and Shah, M. 2021. Dogfight: Detecting drones from drones videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7067--7076

    work page 2021

  4. [4]

    Carion, N.; Massa, F.; Synnaeve, G.; Usunier, N.; Kirillov, A.; and Zagoruyko, S. 2020. End-to-end object detection with transformers. In European conference on computer vision, 213--229. Springer

    2020

  5. [5]

    N.; Fan, A.; Auli, M.; and Grangier, D

    Dauphin, Y. N.; Fan, A.; Auli, M.; and Grangier, D. 2017. Language modeling with gated convolutional networks. In International conference on machine learning, 933--941. PMLR

    work page 2017

  6. [6]

    Ding, J.; Xue, N.; Long, Y.; Xia, G.-S.; and Lu, Q. 2018. Learning RoI transformer for detecting oriented objects in aerial images. arXiv preprint arXiv:1812.00155

    work page Pith review arXiv 2018

  7. [7]

    Ding, X.; Zhang, X.; Ma, N.; Han, J.; Ding, G.; and Sun, J. 2021. Repvgg: Making vgg-style convnets great again. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 13733--13742

    work page 2021

  8. [8]

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

    arXiv 2020

Show all 38 references
  1. [9]

    Du, B.; Huang, Y.; Chen, J.; and Huang, D. 2023. Adaptive sparse convolutional networks with global context enhancement for faster object detection on drone images. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 13435--13444

    2023

  2. [10]

    Du, D.; Qi, Y.; Yu, H.; Yang, Y.; Duan, K.; Li, G.; Zhang, W.; Huang, Q.; and Tian, Q. 2018. The unmanned aerial vehicle benchmark: Object detection and tracking. In Proceedings of the European conference on computer vision (ECCV), 370--386

    2018

  3. [11]

    Duan, K.; Bai, S.; Xie, L.; Qi, H.; Huang, Q.; and Tian, Q. 2019. Centernet: Keypoint triplets for object detection. In Proceedings of the IEEE/CVF international conference on computer vision, 6569--6578

    2019

  4. [12]

    Girshick, R. 2015. Fast r-cnn. In Proceedings of the IEEE international conference on computer vision, 1440--1448

    2015

  5. [13]

    Han, D.; Yun, S.; Heo, B.; and Yoo, Y. 2021. Rethinking channel dimensions for efficient model design. In Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition, 732--741

    2021

  6. [14]

    Hollard , L.; Mohimont , L.; Gaveau , N.; and Steffenel , L.-A. 2024. LeYOLO, New Scalable and Efficient CNN Architecture for Object Detection . arXiv e-prints, arXiv:2406.14239

    2024 arXiv

  7. [15]

    Howard, A.; Sandler, M.; Chu, G.; Chen, L.-C.; Chen, B.; Tan, M.; Wang, W.; Zhu, Y.; Pang, R.; Vasudevan, V.; et al. 2019. Searching for mobilenetv3. In Proceedings of the IEEE/CVF international conference on computer vision, 1314--1324

    2019

  8. [16]

    Huang, Y.; Chen, J.; and Huang, D. 2022. UFPMP-Det: Toward accurate and efficient object detection on drone imagery. In Proceedings of the AAAI conference on artificial intelligence, volume 36, 1026--1033

    2022

  9. [17]

    Jocher, G. 2020. YOLOv5 by Ultralytics

    2020

  10. [18]

    Jocher, G.; Chaurasia, A.; and Qiu, J. 2023. Ultralytics YOLO

    2023

  11. [19]

    Li, C.; Li, L.; Jiang, H.; Weng, K.; Geng, Y.; Li, L.; Ke, Z.; Li, Q.; Cheng, M.; Nie, W.; et al. 2022. YOLOv6: A single-stage object detection framework for industrial applications. arXiv preprint arXiv:2209.02976

    2022 arXiv

  12. [20]

    Lin, T.-Y.; Maire, M.; Belongie, S.; Hays, J.; Perona, P.; Ramanan, D.; Doll \'a r, P.; and Zitnick, C. L. 2014. Microsoft coco: Common objects in context. In Computer Vision--ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 1...

    2014

  13. [21]

    Liu, C.; Gao, G.; Huang, Z.; Hu, Z.; Liu, Q.; and Wang, Y. 2024. YOLC: You Only Look Clusters for Tiny Object Detection in Aerial Images. IEEE Transactions on Intelligent Transportation Systems

    2024

  14. [22]

    Liu, X.; Peng, H.; Zheng, N.; Yang, Y.; Hu, H.; and Yuan, Y. 2023. Efficientvit: Memory efficient vision transformer with cascaded group attention. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 14420--14430

    2023

  15. [23]

    Liu, Z.; Mao, H.; Wu, C.-Y.; Feichtenhofer, C.; Darrell, T.; and Xie, S. 2022. A convnet for the 2020s. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 11976--11986

    2022

  16. [24]

    Meethal, A.; Granger, E.; and Pedersoli, M. 2023. Cascaded Zoom-in Detector for High Resolution Aerial Images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2045--2054

    2023

  17. [25]

    Redmon, J.; Divvala, S.; Girshick, R.; and Farhadi, A. 2016. You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, 779--788

    2016

  18. [26]

    Redmon, J.; and Farhadi, A. 2017. YOLO9000: better, faster, stronger. In Proceedings of the IEEE conference on computer vision and pattern recognition, 7263--7271

    2017

  19. [27]

    Sandler, M.; Howard, A.; Zhu, M.; Zhmoginov, A.; and Chen, L.-C. 2018. Mobilenetv2: Inverted residuals and linear bottlenecks. In Proceedings of the IEEE conference on computer vision and pattern recognition, 4510--4520

    2018

  20. [28]

    Shwartz-Ziv, R.; and Tishby, N. 2017. Opening the black box of deep neural networks via information. arXiv preprint arXiv:1703.00810

    2017 arXiv

  21. [29]

    Tishby, N.; and Zaslavsky, N. 2015. Deep learning and the information bottleneck principle. In 2015 ieee information theory workshop (itw), 1--5. IEEE

    2015

  22. [30]

    Wang, A.; Chen, H.; Lin, Z.; Han, J.; and Ding, G. 2024. Repvit: Revisiting mobile cnn from vit perspective. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 15909--15920

    2024

  23. [31]

    Wang, C.-Y.; Bochkovskiy, A.; and Liao, H.-Y. M. 2023. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7464--7475

    2023

  24. [32]

    Wang, C.-Y.; Yeh, I.-H.; and Liao, H.-Y. M. 2024. YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information. arXiv preprint arXiv:2402.13616

    2024 arXiv

  25. [33]

    Wang, Q.; Wu, B.; Zhu, P.; Li, P.; Zuo, W.; and Hu, Q. 2020. ECA-Net: Efficient channel attention for deep convolutional neural networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 11534--11542

    2020

  26. [34]

    Wei, Z.; Duan, C.; Song, X.; Tian, Y.; and Wang, H. 2020. Amrnet: Chips augmentation in aerial images object detection. arXiv preprint arXiv:2009.07168

    2020 arXiv

  27. [35]

    Yang, C.; Huang, Z.; and Wang, N. 2022. QueryDet: Cascaded sparse query for accelerating high-resolution small object detection. In Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition, 13668--13677

    2022

  28. [36]

    Yang, F.; Fan, H.; Chu, P.; Blasch, E.; and Ling, H. 2019. Clustered object detection in aerial images. In Proceedings of the IEEE/CVF international conference on computer vision, 8311--8320

    2019

  29. [37]

    Ye, H.; Zhang, B.; Chen, T.; Fan, J.; and Wang, B. 2023. Performance-aware approximation of global channel pruning for multitask cnns. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(8): 10267--10284

    2023

  30. [38]

    Zhu, P.; Wen, L.; Du, D.; Bian, X.; Fan, H.; Hu, Q.; and Ling, H. 2021. Detection and tracking meet drones challenge. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(11): 7380--7399

    2021

Pith tools

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