Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

YOLOv11 Optimization for Efficient Resource Utilization

T0 review · 5 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 YOLOv11 can be pruned by object size into six smaller variants that keep accuracy while using fewer resources.

desk verdict A plausible pruning recipe whose efficiency claims are credible, but the 'maintains accuracy' claim is not supported because aggregate mAP on size-skewed data masks the untested separability assumption. read the letter →

arxiv 2412.14790 v3 pith:UARMLKFQ submitted 2024-12-19 cs.CV

Areeg Fahad Rasheed , M. Zarkoosh This is my paper · ORCID
classification cs.CV
keywords YOLOv11objectdetectionnetworkpruningsize-specificmodelcompressioninferencetimeresourceefficiencysizeclassification
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that YOLOv11's three detection heads are specialized by object size, and that pruning the heads and feeding blocks for sizes absent from a dataset yields models that match or slightly beat the original's accuracy while using fewer resources. The authors build six variants, corresponding to small, medium, large, small-medium, medium-large, and small-large, plus a small program that counts object areas in a dataset's labels to pick the right variant. On six datasets, the pruned models keep mAP@50 within about two percentage points of the full model, while reducing model size, GFLOPs, inference time, and, in most cases, power consumption. If this holds, object-detection deployments with a fixed mix of object sizes can replace the full network with a smaller tailored one.

What carries the argument

The load-bearing mechanism is the YOLOv11 head structure: three detection blocks each consuming feature maps at a particular scale (80x80, 40x40, and 20x20), conventionally associated with small, medium, and large objects. The authors label every architectural block from b0 to b22, then for each variant delete the detection head (or heads) for the sizes they want to drop, along with the upstream blocks that feed those heads, reconnecting surviving paths to keep the network a valid feed-forward graph. The object-size classifier program reads label files, bins each bounding-box area into the paper's size categories, and determines which variant matches a dataset's dominant object sizes.

What would settle it

Take a dataset containing objects of all three size classes, train both the full YOLOv11 and one pruned variant (for example YOLOv11-small) on the same data, then compute mAP@50 restricted to small objects only. If the pruned model's small-object mAP falls more than about two points below the full model's, the head-separability assumption is wrong.

Watch

Extended reading notes

Core claim

The central claim is that object-size specialization is built into the YOLOv11 architecture: the three detection heads receive feature maps at different scales (80x80, 40x40, 20x20), so blocks that feed a head for a size range not present in a dataset can be removed without materially hurting detection of the sizes that remain. In experiments across six datasets, each pruned variant's mAP@50 stayed within roughly two percentage points of the full YOLOv11, and four of the six variants actually surpassed it. Model size fell from about 6.3 MB to as low as 3.4 MB, GFLOPs dropped substantially, and inference time decreased in every case.

Load-bearing premise

The argument assumes that the three detection heads of YOLOv11 operate independently by object size, so deleting the blocks that feed the heads for sizes absent from a dataset does not remove features that the remaining heads still need; if the heads share features through the neck, accuracy on the target size will drop.

Editorial extensions

If this is right

  • For deployments where object sizes are roughly fixed, such as aerial imagery, medical cells, or industrial inspection, a pruned variant can replace the full YOLOv11 with no meaningful accuracy loss and a smaller memory footprint.
  • The same pruning recipe can be applied to the larger YOLOv11 variants (s, m, l, x), since the modifications target a shared block structure.
  • The object-size classifier makes model selection automatic: given a dataset's labels, it counts instances per size bin and suggests the smallest architecture whose size coverage fits that distribution.
  • Reduced GFLOPs and faster inference make the pruned models better suited to edge or real-time deployments than the original YOLOv11.
  • Combined variants (sm, ml, sl) cover bimodal size distributions, extending the approach beyond single-size datasets.

Reading between the lines

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

  • The paper's limitation note implies that for scenes with continuously varying object sizes, such as cars on a road growing and shrinking with distance, the full model remains safer; a natural extension would be a size-aware router that switches among pruned variants per image or per frame rather than per dataset.
  • The claim that pruning heads for absent sizes leaves target-size accuracy intact could be tested more directly by measuring per-size mAP (small, medium, large) rather than the aggregate mAP@50 the paper reports.
  • If the head-separability claim transfers across YOLO generations, the same block-removal strategy could be applied to YOLOv8, YOLOv10, or future versions, potentially without retraining from scratch.
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

5 major / 5 minor

Summary. The paper proposes six size-specific pruned variants of YOLOv11n (small, medium, large, sm, ml, sl) by removing blocks in the neck and head that are claimed to serve non-target object sizes, together with a small program that classifies objects in a dataset by size and recommends one of the variants. The variants are evaluated on six datasets and compared with YOLOv11, YOLOv10, and YOLOv8 in terms of recall, precision, mAP@50, model size, GFLOPs, inference time, and power consumption. The central claim is that the pruned models maintain the accuracy of the full YOLOv11 while substantially reducing resource use.

Significance. If the accuracy-preservation claim were established, the proposed variants would be a useful practical contribution for applications with a fixed, known object-size distribution, and the public code and weights would support reproducibility. The efficiency gains are structurally credible: removing detection heads and neck branches necessarily reduces parameter count and computation. However, the current evidence does not support the central accuracy claim because only aggregate mAP on size-skewed datasets is reported, no per-size AP is provided, the experiments use a single seed, and the pruning rationale contains an internal inconsistency about which feature levels detect which object sizes. The paper therefore needs a substantial experimental revision before the central claim can be accepted.

major comments (5)
  1. [§3.1–3.6 and Table 4] The claim that the pruned variants 'maintain accuracy' rests on an untested separability assumption: that the three detection heads and their feeding blocks are functionally separable by object size, so that removing all blocks for non-target sizes leaves target-size accuracy unchanged. No per-size average precision is reported. Table 4 gives only aggregate mAP@50 on datasets that are heavily skewed toward the target size (Table 2: WeedCrop has 15,237/18,693 small objects; Underwater Pipes has 11,683/12,238 large; Brain Tumor has 16,985/21,526 large). Aggregate mAP on such datasets is dominated by the dominant size and cannot reveal whether target-size AP is preserved or whether the model now misses non-target sizes. Please report per-size AP for every dataset/model, or evaluate on a size-balanced benchmark, and provide an ablation that removes one block group at a time.
  2. [§2 and §3] The feature-map semantics used to justify pruning are internally inconsistent and opposite to standard YOLO. Section 2 states that 'the smallest feature sets, such as 20x20, focus on detecting small objects,' whereas in standard YOLO the 20x20 / stride-32 head detects large objects and the 80x80 / stride-8 head detects small objects. Section 3.1 then calls the 'first detection head' the small-object head. Because the choice of which blocks to remove (b14–b16 versus b17–b22) depends on which head corresponds to which object size, an incorrect mapping would remove the wrong heads and invalidate the architectural rationale. Please state explicitly, for each of the three heads, the feature level (P3/P4/P5), stride, and object-size assignment, and reconcile this with the pruning descriptions.
  3. [Abstract and §5.2, Figure 11] The computational-cost values are reported with incorrect units. The abstract says 'the required GFLOPs were reduced from 6.3MB (YOLOv11), 5.7MB (YOLOv10) and 8.1MB (YOLOv8) to just 3.8MB for the large model,' but these numbers appear to be model sizes in megabytes, not GFLOPs. Section 5.2 and Figure 11 also use 'GFLOPS' interchangeably with 'GFLOPs' and describe it as 'giga floating-point operations per second.' Please report the actual GFLOPs per inference for each model in the proper units, and present model size in MB separately, as in Figure 10.
  4. [§5.1, Tables 3 and 4] The accuracy comparison is based on a single seed (Seed 0, Table 3) with no error bars or repeated runs, and the maximum differences in Table 4 are generally under 2%, with some entries equal (e.g., Face Detection recall is 93.10 for both YOLOv11 and YOLOv11-sl). These differences cannot be distinguished from training noise. Please run at least three seeds per condition and report mean and standard deviation, or otherwise provide a significance test, before claiming that the modified models 'maintain' or 'outperform' the original YOLOv11.
  5. [§4.2, Tables 2 and 4, §5.1] There is a factual inconsistency about which model was used for the Brain Tumor dataset. Section 4.2 says 'the YOLOv11-sm modified version is used for the evaluation process of this dataset,' but Table 2 and Table 4 both list YOLOv11-ml for Brain Tumor. Relatedly, Section 5.1 states that 'the original YOLOv11 outperforms the modified versions (YOLOv11-medium and YOLOv11-sm),' while Table 4 shows YOLOv11-sm achieves higher mAP@50 than YOLOv11 on Aerial Airport (93.20 vs. 92.93); the text should refer to YOLOv11-ml. Please correct these inconsistencies, since the model-selection procedure is part of the contribution.
minor comments (5)
  1. [Global] There are several typos and inconsistent terms: 'Limition' in Section 7, 'blcok' in §3.5, 'V alue' in Table 3, 'YOLO10' versus 'YOLOv10' in §5.1, and 'measures the abbreviation measures' in §5.2. A careful proofreading pass is needed.
  2. [§5.2 and Figure 13] The measurement protocol for inference time and power consumption is not described. Please specify the hardware, software versions, batch size used for timing, and how 'Watts per epoch' was measured; this is needed for the resource-efficiency claims to be reproducible.
  3. [References] The Aerial Airport dataset is cited as [7,12] in the text but as [12] in Table 2, and the reference numbering should be aligned across the body, tables, and figures.
  4. [§4.1] The object-size classifier is described only in prose; providing the exact decision rule and an example output would make the model-selection step more transparent and reproducible.
  5. [§5.2] Figure 11 is labeled 'Computational Performance (GFLOPS)' but the text discusses 'GFLOPs'; please use one term consistently and define it as either total operations per inference or operations per second.

Circularity Check

0 steps flagged · score 1.0 of 10

No meaningful circularity: pruning choices and size categories come from external definitions, and the accuracy claim is an empirical comparison rather than a fitted prediction.

full rationale

The paper's derivation chain is not circular. The size thresholds (small area <= 32^2, medium 32^2 < area <= 96^2, large area > 96^2) are taken from an external Ultralytics GitHub issue [31], and the mapping between feature scales and object sizes is attributed to external papers [4, 10]. The six architectures are obtained by removing blocks that the authors associate with non-target sizes; this is an architectural design choice, not a quantity fitted to the evaluation data. The central claim that the modified models 'maintain accuracy' is supported by Table 4, an empirical comparison against YOLOv11, YOLOv10, and YOLOv8, not by an equation that forces the outcome. The object-size classifier merely counts label-box areas and selects a model by the same size ranges; it does not generate the reported mAP values. The only self-citation, reference [25], appears in a background list ('convolutional neural networks (CNNs) [11, 14, 25]') and is not load-bearing for any pruning or accuracy claim. Section 7 further frames the proposal as application-dependent, reinforcing that the accuracy statement is a measured, caveated empirical result. Concerns about the untested separability of detection heads and the absence of per-size AP are evidentiary weaknesses, not circularity: they do not make a prediction equal to its input by construction.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the separability of YOLOv11's heads and the sufficiency of single-seed comparisons. The only hand-picked constants are the object area thresholds borrowed from an external GitHub issue. No new objects or forces are posited.

free parameters (1)
  • Object size thresholds (32^2 and 96^2 pixels) = 32^2 and 96^2
    Hand-chosen boundaries adopted from an Ultralytics GitHub issue (reference [31]) that define small, medium, and large object categories. The six model variants and the classifier program are built on these thresholds, so the central claim depends on them.
assumptions (4)
  • domain assumption YOLOv11's detection heads map cleanly to small, medium, and large objects via the P3, P4, P5 feature maps.
    Invoked in Section 3 when deciding which blocks to remove; the paper's own text contradicts standard YOLO feature-size mapping by stating smaller feature maps detect small objects.
  • domain assumption Removing blocks that feed non-target detection heads leaves target-size detection accuracy unchanged.
    Central architectural assumption behind all six variants; stated in Section 3.1 ('we also eliminated the blocks that fed them with features important for detecting larger sizes').
  • domain assumption Bounding box area in pixels is a valid proxy for the object size categories relevant to detection.
    Used by the object size classifier in Section 4.1; area thresholds may not reflect semantic object size or context.
  • domain assumption Single-seed training with fixed hyperparameters is sufficient to compare model accuracy.
    All runs use Seed 0 and 150 epochs (Table 3); no repeated runs or confidence intervals are reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of YOLOv11 Optimization for Efficient Resource Utilization." pith.science (2026). https://pith.science/paper/UARMLKFQ

@misc{pith2026241214790,
  author       = {Pith},
  title        = {Pith review of: YOLOv11 Optimization for Efficient Resource Utilization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UARMLKFQ}},
  note         = {Machine review of arXiv:2412.14790}
}
read the original abstract

The objective of this research is to optimize the eleventh iteration of You Only Look Once (YOLOv11) by developing size-specific modified versions of the architecture. These modifications involve pruning unnecessary layers and reconfiguring the main architecture of YOLOv11. Each proposed version is tailored to detect objects of specific size ranges, from small to large. To ensure proper model selection based on dataset characteristics, we introduced an object classifier program. This program identifies the most suitable modified version for a given dataset. The proposed models were evaluated on various datasets and compared with the original YOLOv11 and YOLOv8 models. The experimental results highlight significant improvements in computational resource efficiency, with the proposed models maintaining the accuracy of the original YOLOv11. In some cases, the modified versions outperformed the original model regarding detection performance. Furthermore, the proposed models demonstrated reduced model sizes and faster inference times. Models weights and the object size classifier can be found in this repository

Figures

Figures reproduced from arXiv: 2412.14790 by the authors.

Figure 1
Figure 1. Showing the research trends for one-stage [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. YOLOv11 architecture 4 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 5
Figure 5. YOLOv11-Large Architecture for Large Object Detection was reconnected to receive features from block b6, as both utilize the same feature map as input. Addition￾ally, blocks b19, b21, and b22 were renamed to b11, b12, and b13, respectively. The updated architecture of YOLOv11 for large objects is shown in [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: YOLOv11-Medium Architecture for Medium Object Detection [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: YOLOv11-sm Architecture for Small and Medium Object Detection [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: YOLOv11-ml Architecture for Medium and Large Object Detection 3.5 YOLOv11-ml The fifth modified version of YOLOv11, referred to as YOLOv11-ml, is designed to target medium and large objects, with object sizes greater than 322 . This means that any object exceeding this…
Figure 9
Figure 9. Figure 9: Representative samples from the datasets [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: Model Size (MB) [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 11
Figure 11. Figure 11: Computational Performance (GFLOPS) 9 [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 12
Figure 12. Figure 12: Average Inference Time per Image (ms) [PITH_FULL_IMAGE:figures/full_fig_p011_12.png]
Figure 13
Figure 13. Figure 13: Average Power Usage (Watts per epoch) 11 [PITH_FULL_IMAGE:figures/full_fig_p011_13.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Bearded Dragon Activity Recognition Pipeline: An AI-Based Approach to Behavioural Monitoring

    cs.CV 2025-07 conditional novelty 4.0 of 10

    A YOLO plus rule-based pipeline detects bearded dragon basking from video, but hunting detection is limited by poor cricket detection.

Reference graph

Works this paper leans on

38 extracted references · 33 canonical work pages · cited by 1 Pith paper

  1. [1]

    R. 100. brain tumor dataset. https: //universe.roboflow.com/roboflow-100/ brain-tumor-m2pbp, may 2023. visited on 2024-11-27

    work page 2023

  2. [2]

    R. 100. underwater pipes dataset. https: //universe.roboflow.com/roboflow-100/ underwater-pipes-4ng4t, may 2023. visited on 2024-11-27. 12

    work page 2023

  3. [3]

    H. A. Ahmed and E. A. Mohammed. Detection and classification of the osteoarthritis in knee joint using transfer learning with convolutional neural networks (cnns). Iraqi Journal of Science, pages 5058–5071, 2022

    work page 2022

  4. [4]

    M. A. R. Alif. Yolov11 for vehicle detection: Advancements, performance, and applications in intelligent transportation systems. arXiv preprint arXiv:2410.22898, 2024

    arXiv 2024

  5. [5]

    C. M. Badgujar, A. Poulose, and H. Gan. Agri- cultural object detection with you only look once (yolo) algorithm: A bibliometric and systematic literature review. Computers and Electronics in Agriculture, 223:109090, 2024

    work page 2024

  6. [6]

    Bharati and A

    P. Bharati and A. Pramanik. Deep learn- ing techniques—r-cnn to mask r-cnn: a survey. Computational Intelligence in Pattern Recogni- tion: Proceedings of CIPR 2019, pages 657–668, 2020

    work page 2019

  7. [7]

    Ciaglia, F

    F. Ciaglia, F. S. Zuppichini, P. Guerrie, M. Mc- Quade, and J. Solawetz. Roboflow 100: A rich, multi-domain object detection benchmark. arXiv preprint arXiv:2211.13523, 2022

    arXiv 2022

  8. [8]

    L. Du, R. Zhang, and X. Wang. Overview of two- stage object detection algorithms. In Journal of Physics: Conference Series, volume 1544, page 012033. IOP Publishing, 2020

    work page 2020

Show all 38 references
  1. [9]

    Face for small large dataset

    Facedataset. Face for small large dataset. https://universe.roboflow.com/ok-4sjtq/ face-for-small-large , may 2024. visited on 2024-11-27

    2024

  2. [10]

    F. Feng, Y. Hu, W. Li, and F. Yang. Im- proved yolov8 algorithms for small object detec- tion in aerial imagery. Journal of King Saud University-Computer and Information Sciences, 36(6):102113, 2024

    2024

  3. [11]

    Fukushima

    K. Fukushima. Neocognitron: A self-organizing neural network model for a mechanism of pat- tern recognition unaffected by shift in position. Biological cybernetics, 36(4):193–202, 1980

    1980

  4. [12]

    Aerial airport dataset

    GDIT. Aerial airport dataset. https://universe.roboflow.com/gdit/ aerial-airport, may 2024. visited on 2024- 11-27

    2024

  5. [13]

    A. Ghosh. Yolov11 overview. https:// learnopencv.com/yolo11/, 2024. Accessed on November 25, 2024

    2024

  6. [14]

    J. Gu, Z. Wang, J. Kuen, L. Ma, A. Shahroudy, B. Shuai, T. Liu, X. Wang, G. Wang, J. Cai, et al. Recent advances in convolutional neural networks. Pattern recognition, 77:354–377, 2018

    2018

  7. [15]

    R. D. Haameid, B. Q. Al-Abudi, and R. N. Has- san. Automatic object detection, labelling, and localization by camera’s drone system. Iraqi Journal of Science, pages 5008–5023, 2021

    2021

  8. [16]

    K. He, X. Zhang, S. Ren, and J. Sun. Spa- tial pyramid pooling in deep convolutional net- works for visual recognition. IEEE transactions on pattern analysis and machine intelligence , 37(9):1904–1916, 2015

    1904

  9. [17]

    Hussain and R

    M. Hussain and R. Khanam. In-depth review of yolov1 to yolov10 variants for enhanced pho- tovoltaic defect detection. In Solar, volume 4, pages 351–386. MDPI, 2024

    2024

  10. [18]

    Jegham, C

    N. Jegham, C. Y. Koh, M. Abdelatti, and A. Hendawi. Evaluating the evolution of yolo (you only look once) models: A comprehensive benchmark study of yolo11 and its predecessors. arXiv preprint arXiv:2411.00201, 2024

    2024 arXiv

  11. [19]

    Khanam and M

    R. Khanam and M. Hussain. Yolov11: An overview of the key architectural enhancements. arXiv preprint arXiv:2410.17725, 2024

    2024 arXiv

  12. [20]

    Khanam, M

    R. Khanam, M. Hussain, R. Hill, and P. Allen. A comprehensive review of convolutional neural networks for defect detection in industrial appli- cations. IEEE Access, 2024

    2024

  13. [21]

    Kutlu, E

    H. Kutlu, E. Avci, and F. ¨Ozyurt. White blood cells detection and classification based on re- gional convolutional neural networks. Medical hypotheses, 135:109472, 2020

    2020

  14. [22]

    Z. Lin, W. Lin, and F. Jiang. Yolov8-dec: En- hancing brain tumor object detection accuracy in magnetic resonance imaging. Progress in Elec- tromagnetics Research M, 129, 2024

    2024

  15. [23]

    M. Liu, X. Wang, A. Zhou, X. Fu, Y. Ma, and C. Piao. Uav-yolo: Small object detection on unmanned aerial vehicle perspective. Sensors, 20(8):2238, 2020

    2020

  16. [24]

    M. G. Ragab, S. J. Abdulkader, A. Muneer, A. Alqushaibi, E. H. Sumiea, R. Qureshi, S. M. Al-Selwi, and H. Alhussian. A comprehensive systematic review of yolo for medical object de- tection (2018 to 2023). IEEE Access, 2024

    2018

  17. [25]

    A. F. Rasheed and M. Zarkoosh. Unveiling derivatives in deep and convolutional neural net- works: A guide to understanding and optimiza- tion. Authorea Preprints, 2024. 13

    2024

  18. [26]

    J. Redmon. You only look once: Unified, real- time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, 2016

    2016

  19. [27]

    S. Ren, K. He, R. Girshick, and J. Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. IEEE transactions on pattern analysis and machine intelligence , 39(6):1137–1149, 2016

    2016

  20. [28]

    Bccd dataset

    Roboflow. Bccd dataset. https://universe. roboflow.com/joseph-nelson/bccd, aug 2022. visited on 2024-11-27

    2022

  21. [29]

    Sudars, J

    K. Sudars, J. Jasko, I. Namatevs, L. Ozola, and N. Badaukis. Dataset of annotated food crops and weed images for robotic computer vision control. Data in brief, 31:105833, 2020

    2020

  22. [30]

    N. C. Thompson, K. Greenewald, K. Lee, and G. F. Manso. The computational limits of deep learning. arXiv preprint arXiv:2007.05558, 2020

    2007 arXiv

  23. [31]

    Yolov11 classification of ob- ject sizes

    Ultralytics. Yolov11 classification of ob- ject sizes. https://github.com/ultralytics/ ultralytics/issues/8090, 2024. Accessed: 2024-11-26

    2024

  24. [32]

    Yolov11 documentation, 2024

    Ultralytics. Yolov11 documentation, 2024. Ac- cessed: 2024-11-24

    2024

  25. [33]

    Viola and M

    P. Viola and M. Jones. Rapid object detection using a boosted cascade of simple features. In Proceedings of the 2001 IEEE computer society conference on computer vision and pattern recog- nition. CVPR 2001, volume 1, pages I–I. Ieee, 2001

    2001

  26. [34]

    C.-Y. Wang, A. Bochkovskiy, and H.-Y. M. Liao. Scaled-yolov4: Scaling cross stage partial net- work. In Proceedings of the IEEE/cvf confer- ence on computer vision and pattern recognition, pages 13029–13038, 2021

    2021

  27. [35]

    J. Wang, W. Yang, H. Guo, R. Zhang, and G.-S. Xia. Tiny object detection in aerial im- ages. In 2020 25th international conference on pattern recognition (ICPR), pages 3791–3798. IEEE, 2021

    2020

  28. [36]

    X. Wu, D. Sahoo, and S. C. Hoi. Recent ad- vances in deep learning for object detection. Neurocomputing, 396:39–64, 2020

    2020

  29. [37]

    Zhang, X

    Y. Zhang, X. Li, F. Wang, B. Wei, and L. Li. A comprehensive review of one-stage networks for object detection. In 2021 IEEE Interna- tional Conference on Signal Processing, Commu- nications and Computing (ICSPCC), pages 1–6. IEEE, 2021

    2021

  30. [38]

    Z. Zou, K. Chen, Z. Shi, Y. Guo, and J. Ye. Ob- ject detection in 20 years: A survey. Proceedings of the IEEE, 111(3):257–276, 2023. 14

    2023

Pith tools

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