Pith. sign in

REVIEW 4 major objections 4 minor 44 references

A New Hybrid Model of Generative Adversarial Network and You Only Look Once Algorithm for Automatic License-Plate Recognition

T0 review · 4 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

Pith's one-line read A selective GAN deblurring step before YOLOv5 detection raises end-to-end license-plate recognition accuracy from 45.80% to 87.26% on blurred images.

desk verdict Useful engineering report with two released datasets and a gated deblurrer, but the headline 40% gain is measured on in-distribution synthetic blur and should not be trusted for real captures. read the letter →

arxiv 2509.06868 v1 pith:XICCMSOE submitted 2025-09-08 cs.CV

classification cs.CV
keywords automaticlicenseplaterecognitionGANdeblurringYOLOv5objectdetectioncharacterselectivepreprocessingIranianplatesreal-time
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

This paper aims to show that a complete automatic license-plate recognition (ALPR) pipeline can get much better accuracy on blurred images by adding a selective GAN-based deblurring step before a YOLOv5 detector, and that the YOLOv5 stages are fast enough for near-real-time use. The authors report that the deblur step raises end-to-end accuracy from 45.80% to 87.26% on a test set where three-quarters of the images are blurred. They also report 95% license-plate detection accuracy and 97% character-recognition accuracy, at 0.026 seconds per detection stage on a Tesla T4 GPU, and a Raspberry Pi implementation at 0.24 seconds total. If these numbers hold, the main value is a practical recipe: blur only when needed, and let a fast one-stage detector handle plate and character localization.

What carries the argument

The load-bearing mechanism is the selective-deblur module: a Laplacian kernel computes edge intensity, a variance threshold decides whether the image is blurred, and only blurred images pass through the Deblur-GAN, a multi-scale Deep Multi-scale CNN generator (three resolution scales, residual blocks, up-convolution concatenation) with a discriminator. The two YOLOv5 heads then carry localization (one class: license plate) and character recognition (44 classes for Iranian plates), with anchor boxes and output head sizes 18 and 147 respectively.

What would settle it

Take real blurred license-plate images from dashcam or handheld video, run the pipeline with and without the Deblur-GAN, and compare end-to-end accuracy; if the gap is much below the reported 41.46 percentage points, the claim is specific to synthetic blur.

Watch

Extended reading notes

Core claim

The central claim is that a selective deblurring pre-processor based on a multi-scale GAN, gated by a Laplacian-variance blur check, is the difference between a failing and a working ALPR system on blurred captures. Without the deblurrer, the YOLOv5 pipeline recognizes 45.80% of plates in a heavily blurred test set; with it, accuracy rises to 87.26%. The paper further claims that YOLOv5 is a good backbone choice for both license-plate detection and character segmentation+recognition because it detects small objects well and runs in 0.026 s per stage, and that the generated public datasets, including augmented weather conditions and synthetic uniform blur kernels, are representative enough to

Load-bearing premise

The deblurring gain is measured on test images from the same synthetic uniform-blur pipeline used for training, so the assumption that this blur represents real camera motion is what carries the 40% improvement claim.

Editorial extensions

If this is right

  • End-to-end recognition on heavily blurred plates jumps from 45.80% to 87.26% when the selective GAN deblurrer is active.
  • YOLOv5 LPD and CR each detect in 0.026 seconds on a Tesla T4, making the non-blurred path near real-time.
  • The same pipeline reaches 0.24 seconds total on a Raspberry Pi-4B, suggesting portability to dashcam-like devices.
  • Publicly released ALPR and deblur datasets enable others to retrain and benchmark Iranian-plate ALPR systems under varied weather and blur.
  • Against prior Iranian ALPR systems, the reported per-image runtime of 0.052 s is lower than the compared baselines (0.120-0.356 s).

Reading between the lines

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

  • Because the deblurring benefit is measured entirely on images blurred with synthetic uniform kernels, deploying on real camera-shake blur—often non-uniform and multi-source—could yield a smaller gain; a real-blur benchmark would settle this.
  • The 2.052 s deblurring cost suggests the selective gate matters: on mostly-sharp video streams, skipping the GAN keeps the pipeline at 0.052 s, so the threshold could be tuned per camera.
  • The same selective pre-processing pattern—a cheap blur detector followed by an expensive restoration network—could generalize to other small-object recognition tasks such as traffic-sign or pedestrian detection.
  • A cheaper or distilled deblur network might retain most of the accuracy gain while reducing the 2-second penalty on blurred frames, making the selective module usable in closer to real-time.
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

4 major / 4 minor

Summary. The paper proposes an ALPR pipeline consisting of a selective Deblur-GAN preprocessing module, a YOLOv5 detector for license-plate localization (LPD), and a second YOLOv5 detector for character segmentation/recognition (CR). The selection mechanism uses a Laplacian-variance blur check to decide whether deblurring is applied. The authors introduce new Iranian ALPR and deblur datasets, train the components on augmented data, and report YOLOv5 LPD precision of 0.95 and CR precision of 0.976 with 0.026 s per detection stage on a Tesla T4. The central end-to-end claim is that adding the Deblur-GAN raises accuracy from 45.80% to 87.26% on a 300-image test set described as 75% blurred. A Raspberry Pi implementation is also reported.

Significance. The selective-deblur design is sensible and the public release of Iranian ALPR data is a useful community contribution. If the 40% improvement were validated on realistic blurred captures, the paper would demonstrate a practically valuable preprocessing strategy for ALPR. The YOLOv5 speed numbers are attractive for embedded use. However, the headline gain is measured only on synthetic uniform-kernel blur of the same type used for training, on a small test set without uncertainty quantification, and the real-time claim is based on the non-deblurred path. These gaps currently prevent the central claims from being accepted as stated.

major comments (4)
  1. [Table 7 / Section V.A] The 40% improvement from 45.80% to 87.26% is measured on 300 images described only as '75% blurred'. The provenance of these images is not stated, but the only blur-generation procedure in the paper is Section IV: convolving sharp images with uniform OpenCV kernels of size 7x7 to 19x19. The Deblur-GAN was trained on exactly this synthetic degradation. Section III.A.1 itself concedes, citing [22], that such synthetic blur is 'not sufficiently representative, specifically in a multi-source blur or blur kernel with complex patterns'. Therefore the reported gain is an in-distribution result for the trained blur model and does not establish performance on real camera-shake or motion blur. Please evaluate on real blurred captures or at least on non-uniform/spatially-varying blur, and report per-blur-kernel and per-blur-level results.
  2. [Table 7 / Abstract] The central quantitative claim rests on 300 end-to-end images (225 blurred), with no confidence intervals, no repeated runs, and no per-condition breakdown. The difference 45.80% vs 87.26% is an absolute improvement of 41.46 percentage points, which is not 'nearly 40%' in relative terms (relative improvement is ~90%). Please report uncertainties (e.g., Wilson intervals or bootstrap), a significance test, and a breakdown by blur level and by sharp vs blurred subsets. Without this, the headline accuracy gain is not statistically grounded.
  3. [Table 5 vs Table 6] Table 5 states 80/10/10 train/validation/test splits for the YOLO models, but Table 6 reports 3,752 test images for LPD and 1,525 for CR. These do not match 10% of the stated dataset sizes (39,966 and 28,854). The source and composition of these test sets should be described precisely. In addition, the Abstract's '95% and 97% accuracy' figures correspond to the Precision column of Table 6 (0.95 and 0.976), not to an accuracy metric. Please clarify which metric is being reported and, if plate-level accuracy is intended, compute it explicitly.
  4. [Table 7 / Table 8 / Abstract] The real-time claim is based on the non-deblurred path: Table 7 gives 0.052 s end-to-end without deblurring but 2.052 s with Deblur-GAN. Table 8 compares only the 0.052 s figure against prior work, omitting the 2.052 s selective-deblur pipeline that would be invoked for blurred inputs. The paper should report the end-to-end latency of the full selective pipeline, including the blur check and deblurring when triggered, and discuss whether the system remains real-time under the stated 75%-blur test condition.
minor comments (4)
  1. [Section III.B.1] The blur-check threshold is described as a 'carefully selected user-set threshold', but no value, selection procedure, or sensitivity analysis is given. Since the selective behavior depends on it, at least a brief description of its calibration should be added.
  2. [Table 3] Typographical errors and formatting issues: 'Sumsung' should be 'Samsung'; several cells contain 'NA' repeated as 'NANANP'; superscript footnote markers are unexplained. Also, the table reports 'Blur filter size 7×7,9×9,...,19×9' but the text says '7×7 up to 19×19'.
  3. [References] The dataset link is given only as a GitHub URL [36] with no version or DOI; if the data is publicly released, please provide a stable identifier and clarify the licensing.
  4. [Figure 10] The figure caption says 'without/with Deblur-GAN (left/right)', but the ordering is ambiguous from the text. Please make the caption explicit and ensure all subfigures are labeled consistently.

Circularity Check

0 steps flagged · score 1.0 of 10

No definitional circularity; the central claims are empirical. The main caveat is that the headline deblur gain is demonstrated on synthetic uniform-kernel blur of the same type used for training, which is an external-validity concern rather than a circularity.

full rationale

The paper contains no analytic derivation chain whose conclusions reduce to their own assumptions by construction. The YOLOv5 LPD/CR accuracy and timing figures are measured on held-out test splits (Table 6), and the end-to-end comparison in Table 7 directly compares the same system with and without the Deblur-GAN preprocessor on a 300-image test set. The Deblur-GAN is trained on a separately generated Deblur dataset and then evaluated on blurred inputs; unless the test set is secretly drawn from the identical generative pipeline and the evaluation is presented as a prediction on real-world blur, this is a standard train/test protocol. No parameter is fitted to the test set and then renamed a prediction, no uniqueness theorem is imported from the authors, and no ansatz is smuggled in via self-citation. The only self-citation, dataset reference [36], is not load-bearing for the method's validity. The paper itself flags the central limitation in Section III.A.1: 'the synthetic blur datasets so generated are not sufficiently representative, specifically in a multi-source blur or blur kernel with complex patterns scenarios [22].' This directly undermines generalization of the 'nearly 40%' improvement to real non-uniform camera shake and moving-object blur, but that is an external-validity / correctness risk, not a circularity. Accordingly, no circular step is identified.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The ledger is light: no new particles, mediators, or theoretical constructs. The two free parameters are the unreported blur threshold and the synthetic kernel sizes. The four axioms are domain assumptions about synthetic blur representativeness, the Laplacian gate, off-the-shelf architecture behavior, and the Iranian character set. The synthetic-blur-representativeness axiom is the most consequential because it underwrites the headline 40% improvement.

free parameters (2)
  • Laplacian variance blur threshold = not reported
    Hand-selected 'user-set threshold' (Section III.B.1) decides whether the system invokes the 2.052 s GAN deblur path. Its value is never stated, so the selective behavior cannot be replicated or audited.
  • Blur kernel sizes for synthetic blur = 7x7, 9x9, ..., 19x19
    Chosen for dataset generation (Section IV, Table 4); the upper bound is set by computational cost ('the largest size achievable was a 19x19 filter'), and the kernel distribution across the 4051 images is not reported. This is an ad hoc choice for the paper's synthetic blur pipeline.
assumptions (4)
  • domain assumption Convolving sharp images with uniform blur kernels of sizes up to 19x19 produces blur representative of real camera shake and motion blur in ALPR.
    Invoked in Section IV to build the Deblur dataset; the 40% gain in Table 7 is measured on images from this same pipeline, so generalization to real blur is assumed, not tested.
  • domain assumption Laplacian variance below a threshold reliably identifies images where GAN deblurring will help rather than hurt.
    Section III.B.1; no data is shown on the distribution of Laplacian variances for sharp versus blurred images, and the threshold value is unreported.
  • domain assumption YOLOv5 and the Deep Multi-scale CNN (Nah et al., [35]) behave as published, and the authors' training reproduces that behavior.
    The paper relies on off-the-shelf behavior of both architectures with no shipped code or weights; convergence is shown only through the paper's own training curves.
  • domain assumption Iranian license plates are fully covered by 44 character classes.
    Section III.B.3 defines 44 CR classes; the completeness of the set is not justified, though it is plausibly correct for Iranian plates.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A New Hybrid Model of Generative Adversarial Network and You Only Look Once Algorithm for Automatic License-Plate Recognition." pith.science (2026). https://pith.science/paper/XICCMSOE

@misc{pith2026250906868,
  author       = {Pith},
  title        = {Pith review of: A New Hybrid Model of Generative Adversarial Network and You Only Look Once Algorithm for Automatic License-Plate Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XICCMSOE}},
  note         = {Machine review of arXiv:2509.06868}
}
read the original abstract

Automatic License-Plate Recognition (ALPR) plays a pivotal role in Intelligent Transportation Systems (ITS) as a fundamental element of Smart Cities. However, due to its high variability, ALPR faces challenging issues more efficiently addressed by deep learning techniques. In this paper, a selective Generative Adversarial Network (GAN) is proposed for deblurring in the preprocessing step, coupled with the state-of-the-art You-Only-Look-Once (YOLO)v5 object detection architectures for License-Plate Detection (LPD), and the integrated Character Segmentation (CS) and Character Recognition (CR) steps. The selective preprocessing bypasses unnecessary and sometimes counter-productive input manipulations, while YOLOv5 LPD/CS+CR delivers high accuracy and low computing cost. As a result, YOLOv5 achieves a detection time of 0.026 seconds for both LP and CR detection stages, facilitating real-time applications with exceptionally rapid responsiveness. Moreover, the proposed model achieves accuracy rates of 95\% and 97\% in the LPD and CR detection phases, respectively. Furthermore, the inclusion of the Deblur-GAN pre-processor significantly improves detection accuracy by nearly 40\%, especially when encountering blurred License Plates (LPs).To train and test the learning components, we generated and publicly released our blur and ALPR datasets (using Iranian license plates as a use-case), which are more representative of close-to-real-life ad-hoc situations. The findings demonstrate that employing the state-of-the-art YOLO model results in excellent overall precision and detection time, making it well-suited for portable applications. Additionally, integrating the Deblur-GAN model as a preliminary processing step enhances the overall effectiveness of our comprehensive model, particularly when confronted with blurred scenes captured by the camera as input.

Figures

Figures reproduced from arXiv: 2509.06868 by the authors.

Figure 1
Figure 1. FIGURE 1: ALPR system steps [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIGURE 2: Number of the generated data items per grid cell [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIGURE 3: Comparative presentation of different versions of the YOLO structure. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: FIGURE 4: Proposed ALPR system’s architecture. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: FIGURE 5: Deblur-GAN structure [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 7
Figure 7. Figure 7: FIGURE 7: ALPR dataset and its augmentation. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 6
Figure 6. Figure 6: FIGURE 6: LP detection result without (top) and with [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 9
Figure 9. Figure 9: FIGURE 9: Training convergence behaviour of YOLOv3 and YOLOv5 modules. [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: FIGURE 10: CR results without/with Deblur-GAN [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 12
Figure 12. Figure 12: FIGURE 12: Implementation of the proposed ALPR system [PITH_FULL_IMAGE:figures/full_fig_p010_12.png]
Figure 11
Figure 11. Figure 11: FIGURE 11: Output results of the proposed ALPR system. [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 40 canonical work pages

  1. [22]

    Learning to deblur,

    C. J. Schuler, M. Hirsch, S. Harmeling, and B. Scholkopf, “Learning to deblur,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 38, no. 7, pp. 1439–1451, 2016

  2. [1]

    A hybrid KNN-SVM model for Iranian li- cense plate recognition,

    S. S.Tabrizi and N. Cavus, “A hybrid KNN-SVM model for Iranian li- cense plate recognition,”12th International Conference on Application of Fuzzy Systems and Soft Computing, ICAFS, vol. 102, pp. 588–594, 2016

  3. [2]

    A new and robust method for character segmentation and recognition in license plate images,

    A. Sedighi and M. Vafadust, “A new and robust method for character segmentation and recognition in license plate images,”Expert Systems with Applications, vol. 38, pp. 13 497 – 13 504, 2011

  4. [3]

    Representation learning in a deep network for license plate recognition,

    S. Rakhshani, E. Rashedi, and H. Nezamabadi-pour, “Representation learning in a deep network for license plate recognition,”Multimedia Tools and Applications, vol. 79, pp. 13 267 – 13 289, 2020

  5. [4]

    An edge-based color-aided method for license plate detection,

    V . Abolghasemi and A. Ahmadyfard, “An edge-based color-aided method for license plate detection,”Image and Vision Computing, vol. 27, pp. 1134 – 1142, 2009

  6. [5]

    A License Plate Detection Algorithm Using Edge Fea- tures,

    M. A. Mobarhan, A. Shahbahrami, S. Parva, M. N. Asadi, and A. A. Khajekini, “A License Plate Detection Algorithm Using Edge Fea- tures,”Advances in Computer Science, Engineering and Applications, vol. 166, pp. 413 – 421, 2012

  7. [6]

    A vehicle license plate detection method using region and edge based methods,

    M. A. Lalimi, S. Ghofrani, and D. McLernon, “A vehicle license plate detection method using region and edge based methods,”Computers and Electrical Engineering, vol. 39, pp. 834 – 845, 2013

  8. [7]

    A hierarchical algorithm for vehicle license plate localization,

    E. Rashedi and H. Nezamabadi-pour, “A hierarchical algorithm for vehicle license plate localization,”Multimed Tools Appl, vol. 77, pp. 2771 – 2790, 2018

Show all 44 references
  1. [8]

    An Iranian License Plate Recognition System Based on Color Features,

    A. H. Ashtari, M. J. Nordin, and M. Fathy, “An Iranian License Plate Recognition System Based on Color Features,”INTELLIGENT TRANSPORTATION SYSTEMS, vol. 15, pp. 1690 – 1705, 2014

  2. [9]

    Accurate Detection and Recognition of Dirty Vehicle Plate Numbers for High-Speed Appli- cations,

    R. Panahi, Member, and I. Gholampour, “Accurate Detection and Recognition of Dirty Vehicle Plate Numbers for High-Speed Appli- cations,”INTELLIGENT TRANSPORTATION SYSTEMS, vol. 18, pp. 767 – 779, 2017

  3. [10]

    Adaptive enhance- ment and binarization techniques for degraded plate images,

    S. Kashef, H. Nezamabadi-pour, and E. Rashedi, “Adaptive enhance- ment and binarization techniques for degraded plate images,”Multimed Tools Appl, vol. 77, pp. 16 579 – 16 595, 2018

  4. [11]

    Towards end-to-end license plate detection and recognition: A large dataset and baseline,

    Z. Xu, W. Yang, A. Meng, N. Lu, H. Huang, C. Ying, and L. Huang, “Towards end-to-end license plate detection and recognition: A large dataset and baseline,” inComputer Vision – ECCV 2018, 2018, pp. 261–277

  5. [12]

    Automatic License Plate Recognition via sliding-window darknet-YOLO deep learning,

    Hendry and R.-C. Chen, “Automatic License Plate Recognition via sliding-window darknet-YOLO deep learning,”Image and Vision Com- puting, vol. 87, 2019

  6. [13]

    A Real-Time License Plate Detection Method Using a Deep Learning Approach,

    S. Khazaee, A. Tourani, S. Soroori, A. Shahbahrami, and C. Y . Suen, “A Real-Time License Plate Detection Method Using a Deep Learning Approach,”International Conference on Pattern Recognition and Artificial Intelligence, vol. 12068, pp. 425 – 438, 2020

  7. [14]

    A Robust Deep Learning Approach for Automatic Iranian Vehicle License Plate Detection and Recognition for Surveillance Systems,

    A. TOURANI1, A. S. 1, S. SOROORI1, S. KHAZAEE2, and C. Y . SUEN, “A Robust Deep Learning Approach for Automatic Iranian Vehicle License Plate Detection and Recognition for Surveillance Systems,”IEEE Access, vol. 8, pp. 201 317 – 201 330, 2020

  8. [15]

    Real-time license plate detection and recognition using deep convolutional neural networks,

    S. M. Silva and C. R. Jung, “Real-time license plate detection and recognition using deep convolutional neural networks,”J. Vis. Commun. Image R, vol. 71, p. 102773, 2020

  9. [16]

    Multinational License Plate Recognition Using Generalized Character Sequence Detection,

    C. Henry, S. Y . Ahn, and S.-W. Lee, “Multinational License Plate Recognition Using Generalized Character Sequence Detection,”IEEE Access, vol. 8, pp. 35 185 – 35 199, 2020

  10. [17]

    An embedded automatic license plate recognition system using deep learning,

    D. M. F. Izidio, A. P. A. Ferreira, H. R. Medeiros, and E. N. da S. Bar- ros, “An embedded automatic license plate recognition system using deep learning,”Design Automation for Embedded Systems, vol. 24, pp. 23 – 43

  11. [18]

    Deblurring images using a wiener filter,

    P. Biswas, A. Sarkar, and M. Mynuddin, “Deblurring images using a wiener filter,”International Journal of Computer Applications, vol. 109, pp. 36–38, 01 2015

  12. [19]

    Single-image deblurring with neural networks: A comparative survey,

    J. Koh, J. Lee, and S. Yoon, “Single-image deblurring with neural networks: A comparative survey,”Computer Vision and Image Understanding, vol. 203, p. 103134, 2021. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S1077314220301533

  13. [20]

    A new image enhancement and super resolution technique for license plate recognition,

    A. Hamdi, Y . K. Chan, and V . C. Koo, “A new image enhancement and super resolution technique for license plate recognition,” Heliyon, vol. 7, no. 11, p. e08341, 2021. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S2405844021024440

  14. [21]

    Generative adversarial nets,

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial nets,” inAdvances in Neural Information Processing Systems, Z. Ghahramani, M. Welling, C. Cortes, N. Lawrence, and K. Q. Weinberger, Eds., vol. 27...

  15. [23]

    Rich feature hierarchies for accurate object detection and semantic segmentation,

    R. B. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,”CoRR, vol. abs/1311.2524, 2013. [Online]. Available: http://arxiv.org/abs/1311.2524

  16. [24]

    Fast r-cnn,

    R. Girshick, “Fast r-cnn,” in2015 IEEE International Conference on Computer Vision (ICCV), 2015, pp. 1440–1448

  17. [25]

    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,”IEEE Transac- tions on Pattern Analysis and Machine Intelligence, vol. 39, 06 2015

  18. [26]

    Cspnet: A new backbone that can enhance learning capability of cnn,

    C.-Y . Wang, H.-y. Liao, Y .-H. Wu, P.-Y . Chen, J.-W. Hsieh, and I.-H. Yeh, “Cspnet: A new backbone that can enhance learning capability of cnn,” in2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 06 2020, pp. 1571–1580

  19. [27]

    You only look once: Unified, real-time object detection,

    J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once: Unified, real-time object detection,” in2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779– 788

  20. [28]

    Yolo9000: Better, faster, stronger,

    J. Redmon and A. Farhadi, “Yolo9000: Better, faster, stronger,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 07 2017, pp. 6517–6525

  21. [29]

    Yolov3: An incremental improvement,

    ——, “Yolov3: An incremental improvement,”CoRR, vol. abs/1804.02767, 2018. [Online]. Available: http://arxiv.org/abs/1804.02767

  22. [30]

    Yolov4: Optimal speed and accuracy of object detection,

    A. Bochkovskiy, C.-Y . Wang, and H.-Y . M. Liao, “Yolov4: Optimal speed and accuracy of object detection,” 2020

  23. [31]

    Jocheret al., “Yolov5.” [Online]

    G. Jocheret al., “Yolov5.” [Online]. Available: https://github.com/ultralytics/yolov5

  24. [32]

    Feature pyramid networks for object detection,

    T.-Y . Lin, P. Dollar, R. Girshick, K. He, B. Hariharan, and S. Belongie, “Feature pyramid networks for object detection,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 2017

  25. [33]

    Spatial pyramid pooling in deep convolutional networks for visual recognition,

    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, 06 2014

  26. [34]

    Path aggregation network for instance segmentation,

    S. Liu, L. Qi, H. Qin, J. Shi, and J. Jia, “Path aggregation network for instance segmentation,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2018

  27. [35]

    Deep multi-scale convolutional neural network for dynamic scene deblurring,

    S. Nah, T. Kim, and K. M. Lee, “Deep multi-scale convolutional neural network for dynamic scene deblurring,” 07 2017, pp. 257–265

  28. [36]

    Dataset,

    B. Shafizadeh, N. Pourhadi, M. Kelarestaghi, and F. Eshghi, “Dataset,” https://github.com/behnoudshafizadeh/, 2022

  29. [37]

    A. B. Jung, K. Wada, J. Crall, S. Tanaka, J. Graving, C. Reinders, S. Yadav, J. Banerjee, G. Vecsei, A. Kraft, Z. Rui, J. Borovec, C. Vallentin, S. Zhydenko, K. Pfeiffer, B. Cook, I. Fernandez, F.-M. De Rainville, C.-H. Weng, A. Ayala-Acevedo, R. Meudec, M. La- porteet al., “i...

  30. [38]

    Make Sense,

    P. Skalski, “Make Sense,” https://github.com/SkalskiP/make-sense/, 2019

  31. [39]

    A robust real-time automatic license plate recognition based on the yolo detector,

    R. Laroca, E. Severo, L. Zanlorensi, L. Soares de Oliveira, G. Goncalves, W. Schwartz, and D. Menotti, “A robust real-time automatic license plate recognition based on the yolo detector,” 07 2018, pp. 1–10

  32. [40]

    Automatic license plate detection in hazardous condition,

    S. Azam and M. M. Islam, “Automatic license plate detection in hazardous condition,”J. Vis. Commun. Image Represent., vol. 36, pp. 172–186, 2016

  33. [41]

    Zemris License Plate Dataset,

    Zemris, “Zemris License Plate Dataset,” http://www.zemris.fer.hr/projects/LicensePlates/

  34. [42]

    Holistic recognition of low quality license plates by cnn using track annotated data,

    J. Spanhel, J. Sochor, R. Juranek, A. Herout, L. Marsik, and P. Zemcik, “Holistic recognition of low quality license plates by cnn using track annotated data,” in2017 14th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS), 2017, pp. 1–6

  35. [43]

    Robust license plate detection in the wild,

    G.-S. Hsu, A. Ambikapathi, S.-L. Chung, and C.-P. Su, “Robust license plate detection in the wild,” in2017 14th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS), 2017, pp. 1–6

  36. [44]

    Benchmark for license plate character segmentation,

    G. R. Gonçalves, S. P. G. da Silva, D. Menotti, and W. R. Schwartz, “Benchmark for license plate character segmentation,”Journal of Electronic Imaging, vol. 25, no. 5, p. 053034, 2016. [Online]. Available: https://doi.org/10.1117/1.JEI.25.5.053034

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.