Pith. sign in

REVIEW 3 major objections 7 minor 36 references

Detecting Wildfire Flame and Smoke through Edge Computing using Transfer Learning Enhanced Deep Learning Models

T0 review · 3 major / 7 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 Fine-tuning a wildfire detector from a large fire/smoke dataset lifts its accuracy from 45.7% to 79.2% mAP on a small aerial test set.

desk verdict Useful, honest empirical follow-up on transfer learning for CPU-only wildfire detection, but the headline scratch-vs-fine-tune comparison is confounded by different learning rates and the private dataset limits reproducibility. read the letter →

arxiv 2501.08639 v1 pith:DBTHHC3C submitted 2025-01-15 cs.CV eess.IV

Giovanny Vazquez , Shengjie Zhai , Mei Yang This is my paper · ORCID
classification cs.CVeess.IV
keywords wildfiredetectiontransferlearningfine-tuningYOLOv5nobjectedgecomputingAFSEdatasetFASDD
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 argues that transfer learning, specifically fine-tuning a pretrained YOLOv5n detector, is the practical route to accurate aerial wildfire flame-and-smoke detection when only a small annotated dataset is available. On its 282-image Aerial Fire and Smoke Essential (AFSE) target set, fine-tuning from homogeneous Flame and Smoke Detection Dataset (FASDD) weights reaches 79.2% mAP@0.5, whereas 150 epochs of training from scratch reaches 45.7% and even 600 epochs reaches only 69.2%. A heterogeneous source (COCO) helps less than the homogeneous source, a second cascaded transfer stage adds no clear benefit, and transfer learning does not change inference speed, power draw, or energy-delay product on the edge device. The same experiments position YOLOv5n as the fastest lightweight detector on CPU-only hardware, processing images at about twice the rate of newer YOLOv8n and YOLO11n variants.

What carries the argument

The load-bearing mechanism is fine-tuning as a two-step transfer-learning procedure: first train a YOLOv5n detector on a large source dataset (COCO or FASDD), then continue training on AFSE with a lowered learning rate, either freezing a number of layers or leaving all layers unfrozen. The comparison that carries the argument is the unfrozen, homogeneous-source case, because that is where the gain is largest. Supporting machinery includes the AFSE target dataset, the 70/15/15 train/validation/test split, mAP@0.5 with IoU threshold 0.5, stratified 5-fold cross-validation for generalizability, and the normalized energy-delay product for edge efficiency.

What would settle it

Train YOLOv5n with the same FASDD-pretrained fine-tuning recipe and evaluate it on an independently annotated aerial wildfire image set the authors did not use; if the fine-tuned model does not beat a from-scratch model trained for the same number of epochs on that set, the transfer-learning advantage claimed here would not generalize.

Watch

Extended reading notes

Core claim

The central claim is that a one-stage fine-tuning transfer-learning recipe — start from weights trained on a large fire/smoke dataset, then fine-tune on the small aerial target with all layers unfrozen and a low learning rate — turns a lightweight detector into a usable wildfire detector. On the AFSE test split, YOLOv5n fine-tuned from FASDD achieves 79.2% mAP@0.5, with fire AP 70.0% and smoke AP 88.5%, compared with 45.7% mAP for 150-epoch scratch training and 69.2% for 600-epoch scratch training. Fine-tuning from COCO reaches 64.8%, showing that homogeneous source data matters. Cascaded transfer learning, whether through D-Fire or a merged FASDD+D-Fire pretraining stage, lands at 78.0–80.5% mAP, so the extra stage is not a reliable win over the single FASDD stage. The paper also reports that transfer learning leaves edge-computing metrics essentially unchanged, since the architecture and input size are the same.

Load-bearing premise

The AFSE dataset — 282 un-augmented images from an aerial perspective, with its ground-truth boxes — is representative enough of real wildfire scenes that the measured mAP gains will carry over to other aerial imagery.

Editorial extensions

If this is right

  • Teams with small bespoke wildfire datasets can get usable detectors by fine-tuning from an existing large flame/smoke dataset rather than training from scratch.
  • The 600-epoch scratch baseline (69.2%) being below the 150-epoch fine-tuned result (79.2%) means transfer learning can save training time while improving accuracy.
  • Homogeneous pretraining matters: FASDD pretraining beats COCO pretraining by 14.4 mAP points, so source-domain similarity is a first-order choice.
  • Cascaded transfer is not worth the extra training time; merging datasets into a single pretraining stage is at least as good and cheaper.
  • On CPU-only edge hardware, YOLOv5n remains competitive and roughly twice as fast as newer nano variants, making it a reasonable default for UAV deployment.

Reading between the lines

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

  • Because AFSE has only 282 images and one split, the 79.2% figure should be read as a single-split result; an external evaluation on unseen aerial wildfire footage would be needed before treating it as a deployment guarantee.
  • The near-identical FPS and power before and after transfer learning suggest the accuracy gain is free at runtime, but also imply that any future edge gains must come from architecture or quantization, not from the choice of pretrained weights.
  • A natural next test is to apply the same FASDD-to-AFSE fine-tuning recipe to other aerially captured fire datasets and check whether the mAP jump persists; the paper's cross-validation variance suggests smaller variation, but not transfer across datasets.
  • The conclusion that cascaded transfer is unnecessary could be tested by varying the order of source datasets; a larger first-stage set or a closer intermediate domain might still help.
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 / 7 minor

Summary. The paper studies transfer learning (TL) for UAV-based wildfire flame and smoke detection, using the authors' private Aerial Fire and Smoke Essential (AFSE) dataset as the target and either FASDD or COCO as the source dataset. The authors compare training from scratch with fine-tuning for YOLOv5n and other lightweight YOLO variants, also evaluating a cascaded TL scheme, a comparison against several non-YOLO detectors, and edge-computing metrics (FPS, power, energy-delay product) on a Raspberry Pi 5. The headline empirical finding is that fine-tuning YOLOv5n from FASDD pre-trained weights reaches 79.2% mAP@0.5 on the AFSE test split, versus 45.7% for 150 epochs of scratch training and 69.2% for 600 epochs of scratch training. The paper also reports that TL does not materially change edge inference metrics and that YOLOv5n is nearly twice as fast as YOLO8n/YOLO11n on the CPU-only edge device.

Significance. If the central comparison were properly controlled, the result would be practically valuable: it suggests that homogeneous-domain pre-training can substantially compensate for small annotated datasets in aerial wildfire detection, and it identifies YOLOv5n as an attractive model for CPU-only UAV edge devices. The paper includes real power measurements, a range of YOLO versions and non-YOLO detectors, and a 5-fold cross-validation attempt, which are useful strengths. However, the main accuracy claim is currently supported only by experiments in which the training schedule (learning rate, epoch count) is varied together with the initialization, so the reported improvement cannot be uniquely attributed to transfer learning. The edge-computing portion is better controlled and the null result there is reported honestly.

major comments (3)
  1. [IV.A, Table VII, Table VIII]
  2. [IV.B.4, Table XI]
  3. [IV.B.2, Fig. 4]
minor comments (7)
  1. [Abstract] The sentence "TL alone did not benefit from the edge computing metrics evaluated" is garbled; it should read "TL alone did not benefit the edge computing metrics evaluated."
  2. [Table VIII vs Table XI] The YOLOv5n test mAP@0.5 is 79.2% in Table VIII but 79.3% in Table XI for what appears to be the same configuration; please reconcile these values or state that they are from different runs/splits.
  3. [Table X] In the MERGED row, validation mAP values are shown as dashes, yet the text comments on validation results for the merged case; please fill in these values or clarify why they are omitted.
  4. [Fig. 4] The block of numbers preceding Fig. 4 is difficult to parse; it should be formatted as a table with labeled rows for training condition and columns for mean, variance, and standard deviation of APfire, APsmoke, and mAP@0.5.
  5. [Table V] The layout of Table V is confusing because multiple versions are packed into each row (e.g., "5 | 6 8 | 9 10 | 11"); use separate rows or clearer column grouping for each YOLO version.
  6. [General] The AFSE dataset is described as developed in the authors' prior work, but there is no statement about its public availability; if it is private, the reproducibility of the headline numbers is limited, and this should be acknowledged.
  7. [References] Reference [29] is an informal blog post for average precision; consider replacing it with a more standard reference, and verify the formatting of the reference for [8].

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the transfer-learning comparison is an empirical benchmark, and the AFSE self-citation is not load-bearing.

full rationale

The paper's central claims are empirical measurements, not derivations. The mAP@0.5 values in Tables VIII-XI are computed from detector outputs against AFSE ground truth using the standard AP/mAP definitions in Eqns. (1)-(4), and the transfer-learning comparison varies the pretraining source (COCO/FASDD) and frozen-layer count while keeping the same detection architecture and target split. No reported result is constructed from a fitted parameter: the 79.2% mAP figure is a measured test-set number, not a value implied by the fine-tuning recipe by construction. The only self-citation is reference [8], which supplies the AFSE dataset and the authors' earlier transfer-learning exploration; citing it for the dataset's origin is not load-bearing for the current quantitative comparisons, because the tables report new runs and the dataset is an experimental artifact rather than an equation or theorem that could make an output equal to an input. The paper's acknowledged per-model variation of epochs and learning rate (Section IV.A) is a potential experimental confound, especially for the scratch-versus-fine-tune comparison, but a confound is a correctness or design concern, not circularity: it does not make the transfer-learning result true by definition or reduce the mAP numbers to the training schedule. No uniqueness theorem, ansatz, or renamed known result is imported from the authors' prior work. Therefore there are no circular steps.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claims rely on an empirical benchmark whose only target is the authors' AFSE dataset. There are no derived constants; the free parameters are training hyperparameters. The assumptions listed cover dataset representativeness, metric choice, power measurement, and platform representativeness.

free parameters (4)
  • Per-model training epochs = 150, 300, 600 for scratch; 75, 150 for fine-tuning; varied for SOTA models
    Epochs are hand-set to limit overfitting and, for the SOTA comparison, to let each model reach a stable mAP. Headline accuracy numbers depend on these choices.
  • Initial learning rate (lr0) = 0.001 for YOLOv5n fine-tuning; 0.0001 for YOLOv8n/11n fine-tuning
    Set manually per model in Table VII and not derived from any rule.
  • Frozen layers during fine-tuning = 0, 5, 10 for YOLOv5n; 0 for other models
    Frozen-layer count is a hand-swept hyperparameter; the best-reported case uses 0 frozen layers.
  • Batch size per GPU = 16 for YOLO models; 1 to 4 for SOTA models
    Adjusted for memory limits in the SOTA comparison, so training budgets are not identical across models.
assumptions (5)
  • domain assumption AFSE annotations and labels are correct and representative of aerial wildfire scenes.
    All accuracy claims are measured on this 282-image dataset in Section III.A. If labels are wrong or the scenes are not representative, the headline mAP values do not generalize.
  • standard math AP and mAP computed with IoU threshold 0.5 is an appropriate measure of detection quality.
    Section III.C defines AP through the precision-recall integral with IoU=0.5. This is a standard evaluation choice rather than a derived quantity.
  • domain assumption Power measurements from a FNIRSI FNB58 USB tester at 100 Hz capture average inference power accurately.
    Edge power and EDP comparisons in Table XIII and Fig. 7 rely on this instrument and sampling rate, as described in Section III.C.
  • domain assumption A Raspberry Pi 5 CPU-only setup is a representative UAV edge-computing platform.
    The FPS, power, and EDP conclusions are meant to guide UAV deployment, but real drone hardware and thermal conditions may differ from the Table II testbed.
  • domain assumption Starting from public pretrained weights (COCO, FASDD) is a valid baseline for transfer learning comparisons.
    Section III.D assumes that the only meaningful difference between scratch and TL training is the source of the starting weights.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Detecting Wildfire Flame and Smoke through Edge Computing using Transfer Learning Enhanced Deep Learning Models." pith.science (2026). https://pith.science/paper/DBTHHC3C

@misc{pith2026250108639,
  author       = {Pith},
  title        = {Pith review of: Detecting Wildfire Flame and Smoke through Edge Computing using Transfer Learning Enhanced Deep Learning Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DBTHHC3C}},
  note         = {Machine review of arXiv:2501.08639}
}
read the original abstract

Autonomous unmanned aerial vehicles (UAVs) integrated with edge computing capabilities empower real-time data processing directly on the device, dramatically reducing latency in critical scenarios such as wildfire detection. This study underscores Transfer Learning's (TL) significance in boosting the performance of object detectors for identifying wildfire smoke and flames, especially when trained on limited datasets, and investigates the impact TL has on edge computing metrics. With the latter focusing how TL-enhanced You Only Look Once (YOLO) models perform in terms of inference time, power usage, and energy consumption when using edge computing devices. This study utilizes the Aerial Fire and Smoke Essential (AFSE) dataset as the target, with the Flame and Smoke Detection Dataset (FASDD) and the Microsoft Common Objects in Context (COCO) dataset serving as source datasets. We explore a two-stage cascaded TL method, utilizing D-Fire or FASDD as initial stage target datasets and AFSE as the subsequent stage. Through fine-tuning, TL significantly enhances detection precision, achieving up to 79.2% mean Average Precision (mAP@0.5), reduces training time, and increases model generalizability across the AFSE dataset. However, cascaded TL yielded no notable improvements and TL alone did not benefit the edge computing metrics evaluated. Lastly, this work found that YOLOv5n remains a powerful model when lacking hardware acceleration, finding that YOLOv5n can process images nearly twice as fast as its newer counterpart, YOLO11n. Overall, the results affirm TL's role in augmenting the accuracy of object detectors while also illustrating that additional enhancements are needed to improve edge computing performance.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 26 canonical work pages

  1. [1]

    Increasing damages from wildfires warrant investment in wildland fire management,

    C. Crowley, A. Miller, R. Richardson, and J. Malcom, “Increasing damages from wildfires warrant investment in wildland fire management,” U.S. Department of the Interior, R-2023-001, 2023

    work page 2023

  2. [2]

    A wildfire smoke detection system using unmanned aerial vehicle images based on the optimized YOLOv5,

    M. Mukhiddinov, A.B. Abdusalomov, and J. Cho, “A wildfire smoke detection system using unmanned aerial vehicle images based on the optimized YOLOv5,” Sensors, vol. 22, no.23, 2022

    work page 2022

  3. [3]

    A review on early wildfire detection from unmanned aerial vehicles using deep learning-based computer vision algorithms,

    A. Bouguettaya, H. Zarzour, A.M. Taberkit, A. Kechida, “A review on early wildfire detection from unmanned aerial vehicles using deep learning-based computer vision algorithms,” Signal Processing, vol.190, 2022

    work page 2022

  4. [4]

    Near-edge computing aware object detection: a review,

    A. Setyanto, T. B. Sasongko, M. A. Fikri and I. K. Kim, “Near-edge computing aware object detection: a review,” in IEEE Access, vol. 12, pp. 2989-3011, 2024

    work page 2024

  5. [5]

    UAV based cost -effective real -time abnormal event detection using edge computing,

    M.S. Alam, B.V. Natesha, T.S. Ashwin, and R.M.R. Guddeti, “UAV based cost -effective real -time abnormal event detection using edge computing,” Multimed. Tools Appl., vol. 78, pp. 35119–35134, 2019

    work page 2019

  6. [6]

    Challenges in energy- efficient deep neural network training with FPGA,

    Y. Tao, R. Ma, M. -L. Shyu and S. -C. Chen, “Challenges in energy- efficient deep neural network training with FPGA,” 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Seattle, WA, USA, 2020, pp. 1602-1611

    work page 2020

  7. [7]

    YOLOv5 Ultralytics,

    G. Jocher, “YOLOv5 Ultralytics,” Available online: https://github.com/ultralytics/yolov5

    work page

  8. [8]

    Transfer learning enhanced deep learning model for wildfire flame and smoke detection,

    G. Vazquez, S. Zhai and M. Yang, “Transfer learning enhanced deep learning model for wildfire flame and smoke detection,” Proc. Int’l Conf. Smart Applications, Commun. and Netw. (SmartNets) , 2024, Harrisonburg, VA, USA, pp. 1-4

    work page 2024

Show all 36 references
  1. [9]

    FASDD: an open-access 100,000-level flame and smoke detection dataset for deep learning in fire detection,

    M. Wang, L. Jiang, P. Yue, D. Yu, and T. Tuo, “FASDD: an open-access 100,000-level flame and smoke detection dataset for deep learning in fire detection,” Earth Syst. Sci. Data, Mar. 2023

    2023

  2. [10]

    An automatic fire detection system based on deep convolutional neural networks for low - power, resource-constrained devices,

    P.V.A.B de Venâncio, A.C. Lisboa, and A.V. Barbosa, “An automatic fire detection system based on deep convolutional neural networks for low - power, resource-constrained devices,” Neural Comput. & Applic., vol. 34, pp. 15349–15368, 2022

    2022

  3. [11]

    Aerial imagery pile burn detection using deep learning: The FLAME dataset,

    A. Shamsoshoara, et al., “Aerial imagery pile burn detection using deep learning: The FLAME dataset,” Comput. Netw., vol. 193, no. 108001, Jul. 2021

    2021

  4. [12]

    Wildland fire detection and monitoring using a drone-collected rgb/ir image dataset,

    X. Chen, B. Hopkins, H. Wang, L. O’Neill, et al., “Wildland fire detection and monitoring using a drone-collected rgb/ir image dataset, ” IEEE Access, vol. 10, pp. 121301-121317, 2022

    2022

  5. [13]

    The evolution of object detection methods,

    Y. Sun, Z. Sun, and W. Chen, “The evolution of object detection methods,” Engineering Applications of Artificial Intelligence , vol. 133, 2024

    2024

  6. [14]

    Bibliometric analysis of one -stage and two -stage object detection,

    A. Lohia, Kadam, K.D. Kadam, R.R. Joshi, and A.M. Bongale, “Bibliometric analysis of one -stage and two -stage object detection, ” Library Philosophy and Practice (e-journal), 2021

    2021

  7. [15]

    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 Trans.Pattern Analysis and Machine Intelligence, vol. 43, no. 5, pp. 1483-1498, 1 May 2021

    2021

  8. [16]

    Dynamic R-CNN: towards high quality object detection via dynamic training,

    H. Zhang, H. Chang, B. Ma, N. Wang, and X. Chen, “Dynamic R-CNN: towards high quality object detection via dynamic training,” in Proc. 16th Europ. Conf. Comput. Vision (ECCV), 2020

    2020

  9. [17]

    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,” in Proc. Adv. in Neural Inf. Processing Syst., vol. 28, 2015

    2015

  10. [18]

    RTMDet: an empirical study of designing real-time object detectors,

    C. Lyu, W. Zhang, H. Huang, Y. Zhou, Y. Wang, Y. Liu, S. Zhang, and K. Chen, “RTMDet: an empirical study of designing real-time object detectors,” arXiv:2212.07784, 2022

    2022 arXiv

  11. [19]

    TOOD: Task- aligned one-stage object detection,

    C. Feng, Y. Zhong, Y. Gao, M.R. Scott, and W. Huang, “TOOD: Task- aligned one-stage object detection,” arXiv:2108.07755, 2021

    2021 arXiv

  12. [20]

    DINO: DETR with improved denoising anchor boxes for end-to-end object detection,

    H. Zhang, F. Li, S. Liu, L. Zhang, H. Su, J. Zhu, L.M. Ni, and H.-Y. Shum, “DINO: DETR with improved denoising anchor boxes for end-to-end object detection,” arXiv:2203.03605, 2022

    2022 arXiv

  13. [21]

    DAB-DETR: dynamic anchor boxes are better queries for DETR ,

    S. Liu, F. Li, H. Zhang, X. Yang, X. Qi, H. Su, J. Zhu, and L. Zhang , “DAB-DETR: dynamic anchor boxes are better queries for DETR ,” arXiv:2201.12329, 2022

    2022 arXiv

  14. [22]

    DN-DETR: accelerate DETR training by introducing query denoising,

    F. Li, H . Zhang, S. Liu, J. Guo, L.M. Ni, and L . Zhang, “DN-DETR: accelerate DETR training by introducing query denoising,” arXiv:2203.01305, 2022

    2022 arXiv

  15. [23]

    Comparing YOLOv3, YOLOv4 and YOLOv5 for autonomous landing spot detection in faulty UAVs,

    U. Nepal and H. Eslamiat, “Comparing YOLOv3, YOLOv4 and YOLOv5 for autonomous landing spot detection in faulty UAVs, ” Sensors, vol. 22, no. 464, 2022

    2022

  16. [24]

    Assessing the effectiveness of YOLO architectures for smoke and wildfire detection,

    E. Casas, L. Ramos, E. Bendek, and F. Rivas-Echeverria, “Assessing the effectiveness of YOLO architectures for smoke and wildfire detection, ” IEEE Access, vol. 11, pp. 96554-96583, 2023

    2023

  17. [25]

    YOLOv9: learning what you want to learn using programmable gradient information,

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

    2024 arXiv

  18. [26]

    YOLOv10: real-time end-to-end object detection,

    A. Wang , et al., “ YOLOv10: real-time end-to-end object detection,”, arXiv:2405.14458, 2024

    2024 arXiv

  19. [27]

    Evaluating the evolution of YOLO ( you only look once) models: a comprehensive benchmark study of YOLO11 and its predecessors,

    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:2411.00201, 2024

    2024 arXiv

  20. [28]

    Microsoft COCO: common objects in context,

    T.-Y Lin, M. Maire, S. Belongie, L. Bourdev, et al., “Microsoft COCO: common objects in context,” in Proc. 13th ECCV, Aug. 2014

    2014

  21. [29]

    What is average precision in object detection & localization algorithms and how to calculate it?

    A. Anwar, “What is average precision in object detection & localization algorithms and how to calculate it?” Available online: https://towardsdatascience.com/what-is-average-precision-in-object- detection-localization-algorithms-and-how-to-calculate-it-3f330efe697b

  22. [30]

    Chapter 15 - a framework for accelerating bottlenecks in GPU execution with assist warps,

    N. Vijaykumar, et. al., “Chapter 15 - a framework for accelerating bottlenecks in GPU execution with assist warps,” in Advances in GPU Research and Practice, H. Sarbazi-Azad, Ed., Boston, MA, USA: Morgan Kaufmann, 2017, ch. 15, pp. 373-415

    2017

  23. [31]

    Research on the influence of the depth and width of YOLOv5 network structure on taffic signal detection performance,

    W. Chao, F. Jingjing, L. Zhuang, and L. Kuanwei, “Research on the influence of the depth and width of YOLOv5 network structure on taffic signal detection performance,” in Proc. 6th CAA Int. Conf. Veh. Control and Intell. (CVCI), 2022, pp. 1-7

    2022

  24. [32]

    YOLOv6 v3.0: A full-scale reloading,

    C. Li, L. Li, Y. Geng, H. Jiang, M. Cheng, B. Zhang, Z. Ke, X. Xu, and X. Chu, “YOLOv6 v3.0: A full-scale reloading,”, arXiv:2301.05586 , 2023

    2023 arXiv

  25. [33]

    Ultralytics YOLO ,

    G. Jocher, A. Chaurasia, and Q. Jing, “ Ultralytics YOLO ,” Available online: https://github.com/ultralytics/ultralytics

  26. [34]

    Ultralytics YOLO 11,

    G. Jocher and J. Qiu, “Ultralytics YOLO 11,” Available online: https://github.com/ultralytics/ultralytics

  27. [35]

    MMDetection: open mmlab detection toolbox and benchmark,

    K. Chen, J. Wang, J. Pang, Y. Cao, Y. Xiong, X. Li, S. Sun, W. Feng, Z. Liu, J. Xu, et al. 2019. “MMDetection: open mmlab detection toolbox and benchmark,” 2019, arXiv preprint arXiv:1906.07155

    2019 arXiv

  28. [36]

    A gentle introduction to k-fold cross-validation

    J. Brownlee. “A gentle introduction to k-fold cross-validation.” Available online: https://machinelearningmastery.com/k-fold-cross-validation/

Pith tools

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