Pith. sign in

REVIEW 5 major objections 5 minor 32 references

EDNet: Edge-Optimized Small Target Detection in UAV Imagery -- Faster Context Attention, Better Feature Fusion, and Hardware Acceleration

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read EDNet, a YOLOv10-based detector with an extra high-resolution head, cheaper context attention, and a new fusion strategy, claims up to 5.6 mAP@50 gain over YOLOv10 on VisDrone while running in real time on edge devices.

desk verdict EDNet is a competent engineering paper whose headline gains over YOLOv10 are plausible but rest on a comparison table with unverified baseline protocols; still worth refereeing. read the letter →

arxiv 2501.05885 v1 pith:PPLDMJ7E submitted 2025-01-10 cs.CV cs.AIcs.LG

Zhifan Song , Yuan Zhang , Abd Al Rahman M. Abu Ebayyeh This is my paper · ORCID
classification cs.CVcs.AIcs.LG
keywords smallobjectdetectionUAVimageryYOLOv10edgecomputingfeaturefusionattentionmechanismVisDronereal-timeinference
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 proposes EDNet, a detection framework built on YOLOv10, and argues that four modifications—an extra high-resolution detection head, a cross-concatenation feature fusion strategy, a C2f-FCA block with faster context attention, and the WIoUv3 box loss—let one architecture beat all compared state-of-the-art detectors on VisDrone while shrinking parameter counts. The central claim is that EDNet consistently outperforms competitors at every model size, with up to a 5.6 mAP@50 gain over YOLOv10 and a 33.3% mAP@50 for a 1.78M-parameter Tiny variant. The motivation is practical: drone imagery is dominated by tiny targets, and edge devices need low-latency, low-memory inference without post-processing. If the claim holds, EDNet offers a scalable accuracy-efficiency trade-off for real-time UAV detection on phones and embedded boards.

What carries the argument

The load-bearing pieces are four. C2f-FCA is a YOLOv10 C2f variant whose bottleneck blends a partial convolution (from FasterNet) with two pointwise convolutions and a Context Anchor Attention sub-block that uses average pooling, a pointwise convolution, and two depth-wise strip convolutions to approximate large-kernel attention cheaply; this is what the paper credits for better feature extraction and parameter reduction. The XSmall detection head adds a 160x160 feature map (downsampled only twice) so tiny objects retain more pixels. The Cross Concat strategy feeds the SPPF output into the cross-concatenation while keeping the PSA output connected to the first upsampling block, giving the neck multi-scale context. WIoUv3 is a dynamic focusing box-loss that down-weights outlier samples via a non-monotonic focusing coefficient. Together these carry the claimed accuracy gain; the ablation attributes the largest single jump (4.2 mAP@50) to the XSmall head plus CCS, and the final C2f-FCA step both improves accuracy and cuts parameters from 29.9M to 25.5M.

What would settle it

Retrain YOLOv10-M under EDNet's exact protocol (200 epochs, SGD lr 0.01, momentum 0.9, same input resolution and augmentation) and evaluate on the VisDrone validation split; if it reaches or exceeds 47.1% mAP@50, the architecture would not be doing the work the comparison attributes to it.

Watch

Extended reading notes

Core claim

EDNet's core discovery is that a YOLOv10 baseline can be made substantially better at small-target detection in aerial images by adding an XSmall detection head at 160x160 resolution, rerouting the SPPF output through a Cross Concat strategy in the neck, replacing the backbone's C2f blocks with C2f-FCA blocks that combine partial convolution and Context Anchor Attention, and switching the box regression loss to WIoUv3. On the VisDrone 2019-DET-val set, EDNet-M reaches 47.1% mAP@50 versus 41.5% for YOLOv10-M, and EDNet-X reaches 50.2% versus 44.8% for YOLOv10-X, with fewer parameters in several comparisons. The paper attributes the gains mainly to richer small-object features from the extra head and the multi-scale context supplied by the cross-concatenation, while the FCA block reduces parameters and adds speed. It also reports real-time deployment: all seven EDNet variants run at 16 to 55 FPS on an iPhone 12 after INT8/FP16 optimization and neural-engine acceleration.

Load-bearing premise

The central comparison assumes that all baseline models in Table II were trained and evaluated under the same protocol as EDNet (epochs, optimizer, resolution, augmentation, validation), so if any baseline numbers come from different settings, the claimed gains could be artifacts of protocol rather than architecture.

Editorial extensions

If this is right

  • At every size tier in Table II, the EDNet variant beats the corresponding YOLOv3/5/6/8/9/10, RT-DETR, VAMYOLOX, and EdgeYOLO numbers on VisDrone val, so the design recipe is claimed to transfer across model capacities.
  • A 1.78M-parameter Tiny variant reaches 33.3% mAP@50, surpassing all compared N-size YOLO models, which would make Raspberry Pi-class edge targets viable.
  • Because YOLOv10 needs no NMS post-processing, EDNet inherits end-to-end inference; on iPhone 12 the seven variants span 16 to 55 FPS, and on a Raspberry Pi 5 the optimized Tiny model is 3.1 times faster than the raw implementation.
  • The ablation suggests each component contributes: WIoU adds 0.3, the XSmall head adds 3.7, CCS adds 0.5, and C2f-FCA adds 0.4 while removing 4.4M parameters, implying the efficiency story rests mainly on the FCA block.

Reading between the lines

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

  • If the gains are reproducible under identical training protocols, the same combination—extra high-resolution head, cross-concatenation, partial-convolution attention, and wise-IoU loss—could transfer to other YOLO-era detectors or to remote-sensing datasets with tiny objects, though the paper only tests VisDrone.
  • The XSmall head adds a 160x160 detection map, which raises compute; the paper's numbers suggest C2f-FCA compensates, but on lower-end CPUs the net speed benefit may shrink—a testable extension would measure per-component latency on ARM CPUs rather than only on A100 and iPhone.
  • Because the FCA block's strip convolutions approximate large-kernel attention at low cost, the design hints at a general low-parameter route to context modeling for small objects, but the paper does not isolate whether the gain comes from the attention mechanism or from the partial-convolution structure.
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

5 major / 5 minor

Summary. The paper proposes EDNet, a YOLOv10-based detector for small objects in UAV imagery. The main architectural contributions are a C2f-FCA block with a Faster Context Attention mechanism, an additional XSmall detection head, a Cross Concat feature-fusion strategy, and the WIoUv3 bounding-box loss. The authors report seven model sizes, evaluate on VisDrone validation, and claim consistent improvements over YOLOv5/6/8/9/10 and RT-DETR, with up to a 5.6 mAP@50 gain over YOLOv10-M, plus real-time deployment measurements on iPhone, iPad, Mac, and Raspberry Pi. The paper also includes an ablation study and hardware acceleration via CoreML, OpenVINO, TensorRT, and ONNX.

Significance. If the empirical claims hold, the paper would provide a useful accuracy-efficiency trade-off for edge-based UAV detection, with a practical deployment story across multiple hardware platforms. The authors release code and pretrained models, report seven scalable variants, and provide per-component ablations. These are concrete strengths. However, the central comparative claim rests on cross-model comparisons whose training and evaluation protocols are not verified, and the reported numbers lack error bars and suffer from validation-set selection bias. The contribution is therefore plausible but not yet rigorously established.

major comments (5)
  1. [Section III-B and Table II] The paper never states whether the YOLOv5/6/8/9/10, RT-DETR, VAMYOLOX, and EdgeYOLO numbers in Table II were retrained in-house under the protocol described in Section III-B or copied from their source papers. mAP on VisDrone is sensitive to training schedule, input resolution, augmentation, and validation procedure, so the claimed 5.6-point gain over YOLOv10-M and the 'significantly fewer parameters' narrative could partly reflect protocol differences rather than architectural improvements. Please provide per-row provenance, and ideally retrain all baselines under the same protocol; at minimum, clearly mark literature-derived values and discuss the sensitivity of the comparison to protocol differences.
  2. [Section III-B and Table II] All results appear to be single runs with no error bars or multiple seeds. The differences in the N and S tiers are small enough that run-to-run variation could change the ranking (e.g., EDNet-N 33.8 vs YOLOv10-N 32.3; EDNet-S 42.5 vs YOLOv8-S 38.5). Please report mean and standard deviation over at least three independent training runs, or otherwise provide evidence that the observed gaps are statistically meaningful.
  3. [Section III-B and III-C] Section III-B states that 'the best-performing model is selected during the training process,' and Table II reports results on the VisDrone validation split. This means the validation set is used both for checkpoint selection and for the headline comparisons, which introduces selection bias and can overstate generalization performance. Please use a held-out test split for final comparisons, or otherwise account for the selection procedure when reporting validation numbers.
  4. [Section III-C and Table II] The claim of 'significantly fewer parameters' is only true when comparing EDNet to larger YOLOv10 variants. At equal nominal sizes, EDNet has more parameters than YOLOv10 in Table II (e.g., EDNet-S 9.3M vs YOLOv10-S 8.0M; EDNet-M 19.1M vs YOLOv10-M 16.5M; EDNet-L 31.7M vs YOLOv10-L 25.7M; EDNet-X 48.7M vs YOLOv10-X 31.6M). The efficiency narrative should be rephrased to make these cross-size comparisons explicit, or supplemented with a same-parameter comparison.
  5. [Section III-C] The text makes quantitative comparisons to VAMYOLOX-X (47.6% mAP@50, 104.6M parameters) and EdgeYOLO-Tiny (5.5M parameters), yet Table II explicitly excludes these models 'to ensure a fair comparison' because of hardware/software differences. This is internally inconsistent: either include them with clearly stated provenance and protocol, or avoid drawing quantitative superiority conclusions from literature numbers obtained under different training and evaluation conditions.
minor comments (5)
  1. [Section II-A, Eqs. (1)-(3)] The CAA attention equation computes the weighting matrix A, but the manuscript does not show how A is applied to the features before the residual addition; please add the missing multiplication or output definition so the block is fully specified.
  2. [Section II-C, Eq. (7)] Equation (7) is typeset ambiguously: the formula for r and the roles of alpha and delta are unclear as printed. Please rewrite it in a standard one-line form.
  3. [Section III-A] The text says the dataset has 10 target categories but lists only nine: pedestrian, people, bicycle, car, van, tricycle, awning-tricycle, bus, and motorbike. The missing category (truck) should be added.
  4. [Table III] The ablation table would be easier to interpret if it also reported the change in FLOPs or latency for each incremental component, since the text discusses computational overhead but the table only lists parameter counts.
  5. [Section IV] The sentence 'all sizes of EDNet’s efficiency is proved on iPhone' is grammatically awkward and overstates what a latency measurement can prove; please rephrase to 'the efficiency of all EDNet sizes was demonstrated on an iPhone.'

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: EDNet's central claims are empirical comparisons on an external benchmark (VisDrone) against external baselines, and the ablations are standard incremental validations rather than definitional reductions.

full rationale

The paper is an empirical architecture paper, not a derivation. Its central claims — higher mAP@50 than YOLOv10 and other detectors, and real-time edge deployment — are grounded in measurements on VisDrone 2019-DET-val and comparisons in Table II against externally published baselines (YOLOv3/5/6/8/9/10, RT-DETR, VAMYOLOX, EdgeYOLO). The architectural components (C2f-FCA, XSmall head, Cross Concat, WIoU) are described as modifications and validated by the incremental ablation in Table III, where each row is a different model evaluated on the same fixed validation set. No equation in the paper defines the reported mAP in terms of a fitted parameter or an input quantity by construction. The ablation baseline (20.4M params, 43.4 mAP@50) matches the YOLOv10-B entry in Table II, consistent with the authors having trained that baseline rather than defining the final number. There are no load-bearing self-citations, no imported uniqueness theorems, and no fitted input renamed as a prediction. Concerns about whether all Table II baselines were trained under the same protocol, and the framing of the 'fewer parameters' claim against larger rather than same-size models, are legitimate empirical-validity and presentation issues, but they are not circularity. The finding is therefore no significant circularity.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central empirical claims rest on hand-set architecture scales, training settings, and the unstated comparability of external baselines. The main assumptions are that VisDrone validation is a reliable estimate of real drone performance, that baselines were produced under equivalent protocols, and that single-run ablations are stable. No new physical entities are introduced.

free parameters (3)
  • EDNet-Tiny depth scale and channel cap = depth_scale 0.20, max_channels 512
    Hand-set to create the 1.78M-parameter Tiny variant; not derived from any principle. The variant's reported 33.3 mAP@50 depends on these values.
  • Training hyperparameters = 200 epochs, lr 0.01, momentum 0.9, batch size unspecified
    These settings are stated in Section III-B but not swept or justified; they affect all reported mAP values, and batch size is not reported.
  • Input resolution = 640x640
    All training and inference are at 640, which determines the P2/P3/P4/P5 feature-map sizes and the latency figures.
assumptions (3)
  • domain assumption VisDrone validation split is used both for model selection and final benchmark
    The paper selects the best model during training on the validation split and then reports those validation numbers as the main comparison; this assumes no overfitting to the validation set.
  • domain assumption Baseline numbers are comparable to EDNet runs
    Table II mixes numbers from prior papers with presumably in-house runs; the paper does not document the training protocol for each baseline, so comparability is assumed.
  • domain assumption Single-run ablation increments are meaningful
    Each ablation row is a single run with no seeds, so 0.3 to 0.5 mAP differences are treated as real effects.

how reviews work

0 comments
Cite this review

Pith. "Pith review of EDNet: Edge-Optimized Small Target Detection in UAV Imagery -- Faster Context Attention, Better Feature Fusion, and Hardware Acceleration." pith.science (2026). https://pith.science/paper/PPLDMJ7E

@misc{pith2026250105885,
  author       = {Pith},
  title        = {Pith review of: EDNet: Edge-Optimized Small Target Detection in UAV Imagery -- Faster Context Attention, Better Feature Fusion, and Hardware Acceleration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PPLDMJ7E}},
  note         = {Machine review of arXiv:2501.05885}
}
read the original abstract

Detecting small targets in drone imagery is challenging due to low resolution, complex backgrounds, and dynamic scenes. We propose EDNet, a novel edge-target detection framework built on an enhanced YOLOv10 architecture, optimized for real-time applications without post-processing. EDNet incorporates an XSmall detection head and a Cross Concat strategy to improve feature fusion and multi-scale context awareness for detecting tiny targets in diverse environments. Our unique C2f-FCA block employs Faster Context Attention to enhance feature extraction while reducing computational complexity. The WIoU loss function is employed for improved bounding box regression. With seven model sizes ranging from Tiny to XL, EDNet accommodates various deployment environments, enabling local real-time inference and ensuring data privacy. Notably, EDNet achieves up to a 5.6% gain in mAP@50 with significantly fewer parameters. On an iPhone 12, EDNet variants operate at speeds ranging from 16 to 55 FPS, providing a scalable and efficient solution for edge-based object detection in challenging drone imagery. The source code and pre-trained models are available at: https://github.com/zsniko/EDNet.

Figures

Figures reproduced from arXiv: 2501.05885 by the authors.

Figure 1
Figure 1. Comparison with state-of-the-art (SOTA) models for object [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The proposed EDNet framework. The main architecture (backbone-neck-head) is illustrated in the center with a more detailed illustration of [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. The proposed C2f-FCA block with Faster Context Attention bottleneck. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Sample predictions with the 1.78M EDNet-Tiny model under various scenarios. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Performance comparison between EDNet and YOLOv10: (a) mAP gain relative to models of equal or larger size; (b) Parameter reduction [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Inference performance of seven EDNet variants on mobile [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Comparison of inference speed in milliseconds before and after [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Inference results on a sample image with challenging targets on iOS deployment. (a) EDNet-Tiny, (b) EDNet-B. The labels and colors for [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 27 canonical work pages

  1. [1]

    Zhu et al., ”Detection and Tracking Meet Drones Challenge,” in IEEE Transactions on Pattern Analysis and Machine Intel- ligence, vol

    P. Zhu et al., ”Detection and Tracking Meet Drones Challenge,” in IEEE Transactions on Pattern Analysis and Machine Intel- ligence, vol. 44, no. 11, pp. 7380-7399, 1 Nov. 2022

    work page 2022

  2. [2]

    Bisio, H

    I. Bisio, H. Haleem, C. Garibotto, F. Lavagetto and A. Scia- rrone, ”Performance Evaluation and Analysis of Drone-Based Vehicle Detection Techniques From Deep Learning Perspec- tive,” in IEEE Internet of Things Journal, vol. 9, no. 13, pp. 10920-10935, 1 July1, 2022

    work page 2022

  3. [3]

    Redmon, S

    J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, ”You Only Look Once: Unified, Real-Time Object Detection,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV , USA, 2016, pp. 779-788

    work page 2016

  4. [4]

    S. Ren, K. He, R. Girshick and J. Sun, ”Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks” in IEEE Transactions on Pattern Analysis & Ma- chine Intelligence, vol. 39, no. 06, pp. 1137-1149, 2017

    work page 2017

  5. [5]

    Carion, F

    N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, and S. Zagoruyko, ”End-to-End Object Detection with Trans- formers,” in *Computer Vision – ECCV 2020*, A. Vedaldi, H. Bischof, T. Brox, and J. M. Frahm, Eds. Cham, Switzerland: Springer, 2020, pp. 213–229

    work page 2020

  6. [6]

    A. Wang, H. Chen, L. Liu, K. Chen, Z. Lin, J. Han, and G. Ding, ”YOLOv10: Real-Time End-to-End Object Detection,” arXiv:2405.14458, 2024

    arXiv 2024

  7. [7]

    X. Wang, H. Chen, X. Chu and P. Wang, ”AODet: Aerial Object Detection Using Transformers for Foreground Regions,” in IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1-11, 2024, Art no. 4106711. 9 Accepted to IEEE UIC 2024

    work page 2024

  8. [8]

    J. Wang, W. Liu, W. Zhang and B. Liu, ”LV-YOLOv5: A light- weight object detector of Vit on Drone-captured Scenarios,” 2022 16th IEEE International Conference on Signal Processing (ICSP), Beijing, China, 2022, pp. 178-183

    work page 2022

Show all 32 references
  1. [9]

    Dosovitskiy, L

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, ”An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale,” Proc. International Conference on Learn...

    2021

  2. [10]

    C. Hao, H. Zhang, W. Song, F. Liu and E. Wu, ”SliNet: Slicing-Aided Learning for Small Object Detection,” in IEEE Signal Processing Letters, vol. 31, pp. 790-794, 2024

    2024

  3. [11]

    Muzammul, A

    M. Muzammul, A. Algarni, Y . Y . Ghadi and M. Assam, ”Enhancing UA V Aerial Image Analysis: Integrating Advanced SAHI Techniques With Real-Time Detection Models on the VisDrone Dataset,” in IEEE Access, vol. 12, pp. 21621-21633, 2024

    2024

  4. [12]

    T. Shi, Y . Ding and W. Zhu, ”YOLOv5s 2E: Improved YOLOv5s for Aerial Small Target Detection,” in IEEE Access, vol. 11, pp. 80479-80490, 2023

    2023

  5. [13]

    J. An, M. D. Putro, A. Priadana and K. -H. Jo, ”Improved YOLOv5 Network with CBAM for Object Detection Vision Drone,” 2023 IEEE International Conference on Industrial Technology (ICIT), Orlando, FL, USA, 2023, pp. 1-6

    2023

  6. [14]

    Niu and Y

    K. Niu and Y . Yan, ”A Small-Object-Detection Model Based on Improved YOLOv8 for UA V Aerial Images,” 2023 2nd In- ternational Conference on Artificial Intelligence and Intelligent Information Processing (AIIIP), Hangzhou, China, 2023, pp. 57-60

    2023

  7. [15]

    C. Yang, X. Li, H. Zhang, Y . Meng, R. Zhang and R. Yuan, ”UA V Small Target Detection in Complex Scenes Based on Improved YOLOv8s,” 2024 39th Youth Academic Annual Conference of Chinese Association of Automation (Y AC), Dalian, China, 2024, pp. 1791-1798

    2024

  8. [16]

    Y . Xiao, T. Xu, X. Yu, Y . Fang and J. Li, ”A Lightweight Fusion Strategy With Enhanced Interlayer Feature Correlation for Small Object Detection,” in IEEE Transactions on Geo- science and Remote Sensing, vol. 62, pp. 1-11, 2024, Art no. 4708011

    2024

  9. [17]

    Y . Yang, X. Gao, Y . Wang and S. Song, ”V AMYOLOX: An Accurate and Efficient Object Detection Algorithm Based on Visual Attention Mechanism for UA V Optical Sensors,” in IEEE Sensors Journal, vol. 23, no. 11, pp. 11139-11155, 1 June1, 2023

    2023

  10. [18]

    S. Liu, J. Zha, J. Sun, Z. Li and G. Wang, ”EdgeYOLO: An Edge-Real-Time Object Detector,” 2023 42nd Chinese Control Conference (CCC), Tianjin, China, 2023, pp. 7507-7512

    2023

  11. [19]

    Elfwing, E

    S. Elfwing, E. Uchibe, and K. Doya, ”Sigmoid-weighted linear units for neural network function approximation in reinforcement learning,” Neural Networks, vol. 107, pp. 3–11, 2018

    2018

  12. [20]

    Jocher, ”YOLOv5 by Ultralytics,” Zenodo, 2020

    G. Jocher, ”YOLOv5 by Ultralytics,” Zenodo, 2020. [Online]. Available: https://doi.org/10.5281/zenodo.3908559

    2020 doi

  13. [21]

    Jocher, A

    G. Jocher, A. Chaurasia, and J. Qiu, ”Ultralytics YOLO,” GitHub repository, 2023. [Online]. Available: https://github.com/ultralytics/ultralytics

    2023

  14. [22]

    Y . Li, N. Miao, L. Ma, F. Shuang, and X. Huang, ”Trans- former for object detection: Review and benchmark,” Engi- neering Applications of Artificial Intelligence, vol. 126, Part C, p. 107021, 2023

    2023

  15. [23]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, ”Attention is all you need,” In Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS’17). Curran Associates Inc., Red Hook, NY , USA...

  16. [24]

    K. He, X. Zhang, S. Ren, and J. Sun, ”Spatial pyramid pooling in deep convolutional networks for visual recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 37, pp. 1904–1916, 2015

    1904

  17. [25]

    J. Chen, S. Kao, H. He, W. Zhuo, S. Wen, C. Lee, and S. G. Chan, ”Run, Don’t Walk: Chasing Higher FLOPS for Faster Neural Networks,” in 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Vancouver, BC, Canada, 2023 pp. 12021-12031

    2023

  18. [26]

    X. Cai, Q. Lai, Y . Wang, W. Wang, Z. Sun, and Y . Yao, ”Poly Kernel Inception Network for Remote Sensing Detection,” in 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, W A, USA, 2024 pp. 27706- 27716

    2024

  19. [27]

    Z. Tong, Y . Chen, Z. Xu, and R. Yu, ”Wise-IoU: Bounding Box Regression Loss with Dynamic Focusing Mechanism,” arXiv preprint arXiv:2301.10051, 2023

    2023 arXiv

  20. [28]

    Y . Zhao, W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y . Liu, and J. Chen, ”DETRs Beat YOLOs on Real-time Object De- tection,” in 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, W A, USA, 2024 pp. 16965-16974

    2024

  21. [29]

    Ackerman and C

    E. Ackerman and C. Gorman, ”This Drone Uses a Smart- phone for Eyes and a Brain,” IEEE Spectrum, Apr. 9, 2015. [Online]. Available: https://spectrum.ieee.org/this-drone-uses- a-smartphone-for-a-brain

    2015

  22. [30]

    Adarsh, P

    P. Adarsh, P. Rathi and M. Kumar, ”YOLO v3-Tiny: Object Detection and Recognition using one stage improved model,” 2020 6th International Conference on Advanced Computing and Communication Systems (ICACCS), Coimbatore, India, 2020, pp. 687-694

    2020

  23. [31]

    C. Li, L. Li, H. Jiang, K. Weng, Y . Geng, L. Li, Z. Ke, Q. Li, M. Cheng, W. Nie, Y . Li, B. Zhang, Y . Liang, L. Zhou, X. Xu, X. Chu, X. Wei, and X. Wei, ”YOLOv6: A single- stage object detection framework for industrial applications,” arXiv:2209.02976, 2022

    2022 arXiv

  24. [32]

    Wang, I.-H

    C.-Y . Wang, I.-H. Yeh, and H.-Y . M. Liao, ”YOLOv9: Learn- ing what you want to learn using programmable gradient information,” arXiv:2402.13616, 2024. 10

    2024 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.