arxiv: 2604.22506 · v1 · submitted 2026-04-24 · 💻 cs.CV

Recognition: unknown

ICPR 2026 Competition on Low-Resolution License Plate Recognition

Rayson Laroca , Valfride Nascimento , Donggun Kim , Sanghyeok Chung , Subin Bae , Uihwan Seo , Seungsang Oh , Chi M. Phung

show 14 more authors

Minh G. Vo Xingsong Ye Yongkun Du Yuchen Su Zhineng Chen Sunhee Heo Hyangwoo Lee Kihyun Na Khanh V. Vu Nguyen Sang T. Pham Duc N. N. Phung Trong P. Le Vy N. Vo Tran David Menotti

Authors on Pith no claims yet

Pith reviewed 2026-05-08 12:38 UTC · model grok-4.3

classification 💻 cs.CV

keywords low-resolution license plate recognitionLRLPR-26 datasetcomputer vision competitionsurveillance imagingdeep learningbenchmark evaluationimage degradation

0 comments

The pith

The first low-resolution license plate recognition competition yields a top score of 82.13% on real-world data.

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

The paper presents the ICPR 2026 Competition on Low-Resolution License Plate Recognition, which used the LRLPR-26 dataset of 20,000 training and 3,000 test tracks, each containing five low-resolution and five high-resolution images of the same plate. It reports that 99 teams submitted entries, with the winner achieving 82.13% recognition rate and four teams over 80%, underscoring the task's difficulty. The paper also reviews the methods of the top five teams and outlines promising research directions for handling low-quality surveillance images.

Core claim

Organized as the first competition focused on low-resolution license plate recognition with real data, the event demonstrates that despite substantial participation from 269 registered teams across 41 countries, the highest recognition rate on the blind test set reached only 82.13%.

What carries the argument

The LRLPR-26 dataset and its track-based evaluation protocol, which pairs multiple low- and high-resolution images per license plate to test methods under operational degradation.

If this is right

Methods that effectively fuse information from the five low-resolution frames per track performed best.
Handling compression artifacts and adverse conditions remains a key challenge for further improvements.
The summary of top approaches highlights the use of deep learning models adapted for low-quality inputs.
Continued progress will likely require new techniques beyond current state-of-the-art to close the gap to 100% accuracy.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

This benchmark may encourage the creation of more robust models that perform well across different surveillance setups.
Future work could explore integrating the high-resolution images more directly into test-time inference if available.
The high number of participants indicates growing practical interest in improving automatic plate reading for security applications.

Load-bearing premise

The LRLPR-26 dataset and evaluation protocol accurately capture the distribution of real-world low-resolution license plate images encountered in operational surveillance scenarios.

What would settle it

Re-evaluating the winning methods on license plate images from a different set of cameras or locations that were not represented in the competition data would show whether the 82.13% rate generalizes or drops.

Figures

Figures reproduced from arXiv: 2604.22506 by Chi M. Phung, David Menotti, Donggun Kim, Duc N. N. Phung, Hyangwoo Lee, Khanh V. Vu Nguyen, Kihyun Na, Minh G. Vo, Rayson Laroca, Sanghyeok Chung, Sang T. Pham, Seungsang Oh, Subin Bae, Sunhee Heo, Trong P. Le, Uihwan Seo, Valfride Nascimento, Vy N. Vo Tran, Xingsong Ye, Yongkun Du, Yuchen Su, Zhineng Chen.

**Figure 1.** Figure 1: Examples of tracks from the training data. Each row corresponds to a complete track, showing five consecutive low-resolution (LR) images on the left and five consecutive high-resolution (HR) captures of the same LP on the right. The original videos (before LP cropping) were acquired with a rolling-shutter camera installed at the Federal University of Paraná, in Curitiba, Brazil, under view at source ↗

**Figure 2.** Figure 2: Representative full-frame images (i.e., before LP detection and cropping) used to construct the LRLPR-26 dataset. The top row corresponds to Scenario A, whereas the bottom row corresponds to Scenario B, illustrating variations in vehicle categories and environmental conditions, including daylight, rain, and nighttime. The cropped LPs were obtained from video sequences of vehicles entering and leaving the r… view at source ↗

**Figure 3.** Figure 3: Overview of the dataset construction pipeline. LPs are first detected using YOLOv11 [60] and then tracked across frames with BoT-SORT [1]. For each vehicle, patches extracted from frames farther from the camera are used as lowresolution (LR) samples, whereas patches from frames closer to the camera are used as high-resolution (HR) samples. The final LP transcription is then obtained semiautomatically by… view at source ↗

**Figure 4.** Figure 4: Recognition Rate achieved by all teams with valid submissions in the Blind Test Phase, sorted by final rank. The dashed horizontal line indicates the mean Recognition Rate across all participants, while the top 20 teams are highlighted in blue. Six teams (ranked 94th to 99th) achieved Recognition Rates at or near 0% and are therefore not visually distinguishable at this scale view at source ↗

**Figure 5.** Figure 5: Recognition Rate versus Confidence Gap for all teams with valid submissions in the Blind Test Phase. The dashed vertical line indicates the Recognition Rate required to enter the top 20, and colors encode the final rank of each team. ture, use of multiple frames, validation protocol, and reliance on external public datasets. 1 st Place The first-place team (DLmath) from Korea University proposed a teacher… view at source ↗

**Figure 6.** Figure 6: Overview of the teacher-student framework proposed by the winning team (DLmath). The student branch processes low-resolution (LR) inputs while the teacher branch, updated via Exponential Moving Average (EMA), receives downsampled high-resolution (HR) images to guide the student’s learning view at source ↗

**Figure 7.** Figure 7: Overall ensemble architecture of the 2nd-place team (AIO_JiangnamCoffee). Five input frames are processed through a shared STN and SE-ResNet34 backbone and fed into four model variants that differ in Transformer encoder depth and the use of auxiliary losses (OHEM-CTC and length penalty). The models were trained from scratch for 30 epochs using AdamW [38] and a one-cycle learning rate schedule. To effective… view at source ↗

**Figure 8.** Figure 8: Overview of the pipeline proposed by the 4th-place team (CAP2 ). It comprises MF-LPR2 -based preprocessing, dual-stream recognition with feature-level and imagelevel branches, and a two-stage position-wise character ensemble for final prediction. All models were trained exclusively on the official training data, with 10% held out for validation, using AdamW [38] with CosineWarmRestart or CosineAnnealing s… view at source ↗

**Figure 9.** Figure 9: Overview of the pipeline proposed by the 5th-place team (UIT-MeoBeo). Native LR images and synthetic LR samples generated from HR license plates were used to train a ViT-based multi-frame recognizer with dual heads for character and layout estimation. During inference, quality-weighted constrained decoding and layout-specific specialist routing were used to obtain the final prediction. Key design choices i… view at source ↗

read the original abstract

Low-Resolution License Plate Recognition (LRLPR) remains a challenging problem in real-world surveillance scenarios, where long capture distances, compression artifacts, and adverse imaging conditions can severely degrade license plate legibility. To promote progress in this area, we organized the ICPR 2026 Competition on Low-Resolution License Plate Recognition, the first competition specifically dedicated to LRLPR using real low-quality data collected under operationally relevant conditions. The competition was based on the LRLPR-26 dataset, which comprises 20,000 training tracks and 3,000 test tracks; each training track contains five low-resolution and five high-resolution images of the same license plate. Notably, a total of 269 teams from 41 countries registered for the competition, and 99 teams submitted valid entries in the Blind Test Phase. The winning team achieved a Recognition Rate of 82.13%, and four teams surpassed the 80% mark, highlighting both the high level of competition at the top of the leaderboard and the continued difficulty of the task. In addition to presenting the competition design, evaluation protocol, and main results, this paper summarizes the methods adopted by the top-5 teams and discusses current trends and promising directions for future research on LRLPR. The competition webpage is available at https://icpr26lrlpr.github.io/

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This is a straightforward competition report that supplies a new real-world dataset and verifiable leaderboard numbers for low-resolution license plate recognition.

read the letter

The paper reports on the ICPR 2026 competition for low-resolution license plate recognition. It introduces the LRLPR-26 dataset and gives the results from 99 valid submissions out of 269 registered teams from 41 countries. The top score reached 82.13 percent, with four teams above 80 percent. That is the core factual content here. The dataset includes 20,000 training tracks, each with five low-resolution and five high-resolution images of the same plate, plus a 3,000-track blind test set. The paper also summarizes the methods used by the top five teams and notes some trends such as super-resolution steps or targeted augmentations. This setup is new in the sense that no prior public competition focused specifically on this narrow surveillance task with paired low- and high-resolution tracks. The reporting is direct and the numbers are presented as direct outcomes of the event, which makes them easy to check. The evaluation protocol is described clearly enough that others could replicate the scoring. The soft spots are limited and typical for this type of paper. The main value is the benchmark itself rather than any new algorithm or theoretical insight, so the novelty sits at the level of providing a shared testbed. Details on exactly how the low-resolution images were captured and verified against operational conditions are light, which leaves some room for doubt about how closely the data matches real deployments. The remark on continued difficulty is simply the observation that scores stayed well below 100 percent and does not rest on extra assumptions. This paper is for researchers working on license plate recognition under poor imaging conditions or anyone who needs a public baseline for low-resolution text recognition in surveillance. A reader who wants concrete numbers to compare against or a starting dataset for their own experiments will find it directly useful. It deserves a serious referee because the claims are factual, the data is public, and the participation statistics add credibility. I would recommend sending it to peer review for a competition track or similar venue, with attention to expanding the dataset collection description.

Referee Report

0 major / 1 minor

Summary. The manuscript reports on the organization of the ICPR 2026 Competition on Low-Resolution License Plate Recognition. It introduces the LRLPR-26 dataset (20,000 training tracks each containing five low-resolution and five high-resolution images of the same plate, plus 3,000 test tracks), documents participation (269 teams from 41 countries registered, 99 valid blind-test submissions), states that the winning team reached a recognition rate of 82.13% with four teams above 80%, summarizes the methods of the top-5 teams, and discusses trends and future directions for LRLPR.

Significance. If the reported facts hold, the paper supplies a concrete, publicly documented benchmark and dataset for low-resolution license plate recognition under operationally relevant conditions. The verifiable participation statistics, submission counts, and leaderboard results provide a clear snapshot of current performance levels and field interest. Credit is given for the factual, non-derivational reporting style and for releasing the competition webpage, which together offer a reusable reference for researchers working on surveillance imagery degraded by distance, compression, and adverse conditions.

minor comments (1)

[Abstract and §4] The abstract and results section could explicitly state the precise definition of the Recognition Rate metric (e.g., character-level accuracy threshold or full-plate match) to make the 82.13% figure immediately interpretable without consulting the competition webpage.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive review, accurate summary of the manuscript, and recommendation to accept. The report correctly identifies the key elements of the ICPR 2026 LRLPR competition, including the dataset design, participation statistics, results, and future directions.

Circularity Check

0 steps flagged

No significant circularity: purely descriptive competition report

full rationale

The paper is a factual competition summary documenting dataset construction (LRLPR-26 with 20k training and 3k test tracks), registration statistics (269 teams, 99 submissions), evaluation protocol, and observed leaderboard outcomes (top RR 82.13%). No derivations, equations, fitted parameters, predictions, or modeling steps exist. Claims rest on direct event results rather than self-referential definitions or unverified self-citations. The interpretive note on task difficulty follows immediately from the reported scores without circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is an empirical competition report with no theoretical derivations, fitted parameters, or new postulated entities. It relies on standard recognition-rate metrics and the assumption that the collected tracks represent operational conditions.

pith-pipeline@v0.9.0 · 5640 in / 1111 out tokens · 54047 ms · 2026-05-08T12:38:56.880417+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

70 extracted references · 48 canonical work pages · 1 internal anchor

[1]

Fleet, and Mohammad Norouzi

Aharon, N., Orfaig, R., Bobrovsky, B.Z.: BoT-SORT: Robust associations multi- pedestrian tracking. arXiv preprint arXiv:2104.07636 pp. 1–13 (2022)

work page arXiv 2022
[2]

In: IEEE/CVF International Conference on Computer Vision (ICCV)

Baek, J., Kim, G., Lee, J., Park, S., Han, D., Yun, S., Oh, S.J., Lee, H.: What is wrong with scene text recognition model comparisons? dataset and model analysis. In: IEEE/CVF International Conference on Computer Vision (ICCV). pp. 4714– 4722 (2019).https://doi.org/10.1109/ICCV.2019.00481

work page doi:10.1109/iccv.2019.00481 2019
[3]

In: European Conference on Computer Vision (ECCV)

Bautista, D., Atienza, R.: Scene text recognition with permuted autoregressive sequence models. In: European Conference on Computer Vision (ECCV). pp. 178– 196 (2022).https://doi.org/10.1007/978-3-031-19815-1_11

work page doi:10.1007/978-3-031-19815-1_11 2022
[4]

In: European Conference on Computer Vision (ECCV)

Carion, N., Massa, F., Synnaeve, G., Usunier, N., Kirillov, A., Zagoruyko, S.: End-to-end object detection with transformers. In: European Conference on Computer Vision (ECCV). pp. 213–229 (2020).https://doi.org/10.1007/ 978-3-030-58452-8_13

2020
[5]

IEEE Transactions on Pattern Analysis and Machine Intelligence48(3), 2676–2694 (2026).https://doi.org/10

Chen, X., Wang, X., Zhang, W., Kong, X., Qiao, Y., Zhou, J., Dong, C.: HAT: Hybrid attention transformer for image restoration. IEEE Transactions on Pattern Analysis and Machine Intelligence48(3), 2676–2694 (2026).https://doi.org/10. 1109/TPAMI.2025.3628275

work page arXiv 2026
[6]

Ev-flying: An event-based dataset for in-the-wild recognition of flying objects

Chen, Z., et al.: NTIRE 2025 challenge on image super-resolution (×4): Methods and results. In: IEEE/CVF Conference on Computer Vision and Pattern Recog- nition Workshops (CVPRW). pp. 1516–1526 (2025).https://doi.org/10.1109/ CVPRW67362.2025.00141

work page arXiv 2025
[7]

In: International Joint Conference on Neural Networks (IJCNN)

Coluccia, A., Fascista, A., Dimou, A., Zarpalas, D., Sommer, L., Schumann, A., Mele, E.: The drone-vs-bird detection grand challenge at IJCNN 2025. In: International Joint Conference on Neural Networks (IJCNN). pp. 1–8 (2025). https://doi.org/10.1109/IJCNN64981.2025.11228314

work page doi:10.1109/ijcnn64981.2025.11228314 2025
[8]

Multimedia Tools and Applications82(6), 9243–9275 (2023).https://doi.org/10.1007/ s11042-022-13644-y

Diwan, T., Anirudh, G., Tembhurne, J.V.: Object detection using YOLO: challenges, architectural successors, datasets and applications. Multimedia Tools and Applications82(6), 9243–9275 (2023).https://doi.org/10.1007/ s11042-022-13644-y

2023
[9]

In: International Joint Confer- ence on Artificial Intelligence (IJCAI)

Du, Y., Chen, Z., Jia, C., Yin, X., Zheng, T., Li, C., Du, Y., Jiang, Y.G.: SVTR: Scene text recognition with a single visual model. In: International Joint Confer- ence on Artificial Intelligence (IJCAI). pp. 884–890 (2022).https://doi.org/10. 24963/ijcai.2022/124

2022
[10]

In: IEEE/CVF International Conference on Computer Vision (ICCV)

Du, Y., Chen, Z., Xie, H., Jia, C., Jiang, Y.G.: SVTRv2: CTC beats encoder- decoder models in scene text recognition. In: IEEE/CVF International Conference on Computer Vision (ICCV). pp. 20147–20156 (Oct 2025)

2025
[11]

In: AAAI Conference on Artificial Intelli- gence

Du, Y., Zhao, M., Fan, S., Chen, Z., Jia, C., Jiang, Y.G.: MDiff4STR: Mask dif- fusion model for scene text recognition. In: AAAI Conference on Artificial Intelli- gence. pp. 3705–3713 (Mar 2026).https://doi.org/10.1609/aaai.v40i5.37370

work page doi:10.1609/aaai.v40i5.37370 2026
[12]

In: ICPR 2026 Competition on Low-Resolution License Plate Recognition 17 Conference on Graphics, Patterns and Images (SIBGRAPI)

Gonçalves, G.R., Diniz, M.A., Laroca, R., Menotti, D., Schwartz, W.R.: Real- time automatic license plate recognition through deep multi-task networks. In: ICPR 2026 Competition on Low-Resolution License Plate Recognition 17 Conference on Graphics, Patterns and Images (SIBGRAPI). pp. 110–117 (Oct 2018).https://doi.org/10.1109/SIBGRAPI.2018.00021

work page doi:10.1109/sibgrapi.2018.00021 2026
[13]

In: Iberoamerican Congress on Pattern Recognition (CIARP)

Gonçalves, G.R., Diniz, M.A., Laroca, R., Menotti, D., Schwartz, W.R.: Multi-task learning for low-resolution license plate recognition. In: Iberoamerican Congress on Pattern Recognition (CIARP). pp. 251–261 (Oct 2019).https://doi.org/10. 1007/978-3-030-33904-3_23

2019
[14]

In: International Joint Conference on Artificial Intelligence (IJCAI)

Gong, H., Feng, Y., Zhang, Z., Hou, X., Liu, J., Huang, S., Liu, H.: A dataset and model for realistic license plate deblurring. In: International Joint Conference on Artificial Intelligence (IJCAI). pp. 1–9 (2024).https://doi.org/10.24963/ ijcai.2024/86

2024
[15]

In: IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR)

Gong, H., Zhang, Z., Feng, Y., Nguyen, A., Liu, H.: LP-Diff: Towards improved restoration of real-world degraded license plate. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 17831–17840 (2025). https://doi.org/10.1109/CVPR52734.2025.01661

work page doi:10.1109/cvpr52734.2025.01661 2025
[16]

In: International Conference on Machine Learning (ICML)

Graves, A., Fernández, S., Gomez, F., Schmidhuber, J.: Connectionist temporal classification: Labelling unsegmented sequence data with recurrent neural net- works. In: International Conference on Machine Learning (ICML). pp. 369–376 (2006).https://doi.org/10.1145/1143844.1143891

work page doi:10.1145/1143844.1143891 2006
[17]

In: IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR)

Guo, H., Guo, Y., Zha, Y., Zhang, Y., Li, W., Dai, T., Xia, S.T., Li, Y.: Mam- baIRv2: Attentive state space restoration. In: IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR). pp. 28124–28133 (2025).https: //doi.org/10.1109/CVPR52734.2025.02619

work page doi:10.1109/cvpr52734.2025.02619 2025
[18]

In: European Conference on Computer Vision (ECCV)

Guo, H., Li, J., Dai, T., Ouyang, Z., Ren, X., Xia, S.T.: MambaIR: A simple baseline for image restoration with state-space model. In: European Conference on Computer Vision (ECCV). pp. 222–241 (2025).https://doi.org/10.1007/ 978-3-031-72649-1_13

2025
[19]

In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

He, T., Zhang, Z., Zhang, H., Zhang, Z., Xie, J., Li, M.: Bag of tricks for image classification with convolutional neural networks. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 558–567 (2019).https: //doi.org/10.1109/CVPR.2019.00065

work page doi:10.1109/cvpr.2019.00065 2019
[20]

IEEE Transactions on Vehicular Technology62(2), 552–561 (Feb 2013).https: //doi.org/10.1109/TVT.2012.2226218

Hsu, G.S., Chen, J.C., Chung, Y.Z.: Application-oriented license plate recognition. IEEE Transactions on Vehicular Technology62(2), 552–561 (Feb 2013).https: //doi.org/10.1109/TVT.2012.2226218

work page doi:10.1109/tvt.2012.2226218 2013
[21]

In: IEEE/CVF Con- ference on Computer Vision and Pattern Recognition

Hu, J., Shen, L., Sun, G.: Squeeze-and-excitation networks. In: IEEE/CVF Con- ference on Computer Vision and Pattern Recognition. pp. 7132–7141 (2018). https://doi.org/10.1109/CVPR.2018.00745

work page doi:10.1109/cvpr.2018.00745 2018
[22]

Hugging Face model hub (2025), timm-remapped image-encoder-only variant

Hugging Face: Model card and pretrained weights for pe-core-l-14-336. Hugging Face model hub (2025), timm-remapped image-encoder-only variant

2025
[23]

IEEE Access13, 145387–145415 (2025).https://doi.org/ 10.1109/ACCESS.2025.3598971

Ismail, A., Mehri, M., Sahbani, A., Essoukri Ben Amara, N.: Automatic license plate recognition in in-the-wild scenarios: A comprehensive review, open issues, and future directions. IEEE Access13, 145387–145415 (2025).https://doi.org/ 10.1109/ACCESS.2025.3598971

work page doi:10.1109/access.2025.3598971 2025
[24]

In: International Conference on Neural Information Processing Systems (NeurIPS)

Jaderberg, M., Simonyan, K., Zisserman, A., Kavukcuoglu, K.: Spatial trans- former networks. In: International Conference on Neural Information Processing Systems (NeurIPS). pp. 2017–2025 (2015)

2017
[25]

In: 2023 IEEE/CVF International Conference on Computer Vision (ICCV)

Jiang, Q., Wang, J., Peng, D., Liu, C., Jin, L.: Revisiting scene text recogni- tion: A data perspective. In: IEEE/CVF International Conference on Computer Vision (ICCV). pp. 20486–20497 (2023).https://doi.org/10.1109/ICCV51070. 2023.01878 18 R. Laroca et al

work page doi:10.1109/iccv51070 2023
[26]

Khalatbarisoltani, L

Ke, X., Zeng, G., Guo, W.: An ultra-fast automatic license plate recognition ap- proach for unconstrained scenarios. IEEE Transactions on Intelligent Transporta- tion Systems24(5), 5172–5185 (2023).https://doi.org/10.1109/TITS.2023. 3237581

work page doi:10.1109/tits.2023 2023
[27]

Computer Vision and Image Understanding238, 103879 (2024).https://doi.org/10.1016/ j.cviu.2023.103879

Kim, D., Kim, J., Park, E.: AFA-Net: Adaptive feature attention network in image deblurring and super-resolution for improving license plate recognition. Computer Vision and Image Understanding238, 103879 (2024).https://doi.org/10.1016/ j.cviu.2023.103879

work page arXiv 2024
[28]

V., Lucio , D

Laroca, R., Cardoso, E.V., Lucio, D.R., Estevam, V., Menotti, D.: On the cross- dataset generalization in license plate recognition. In: International Conference on Computer Vision Theory and Applications (VISAPP). pp. 166–178 (Feb 2022). https://doi.org/10.5220/0010846800003124

work page doi:10.5220/0010846800003124 2022
[29]

Latent Space Alignment for AI -Native MIMO Semantic Communications,

Laroca, R., dos Santos, M., Menotti, D.: Improving small drone detection through multi-scale processing and data augmentation. In: International Joint Conference on Neural Networks (IJCNN). pp. 1–8 (June 2025).https://doi.org/10.1109/ IJCNN64981.2025.11227421

work page arXiv 2025
[30]

In: In- ternational Joint Conference on Neural Networks (IJCNN)

Laroca, R., Estevam, V., Britto Jr., A.S., Minetto, R., Menotti, D.: Do we train on test data? The impact of near-duplicates on license plate recognition. In: In- ternational Joint Conference on Neural Networks (IJCNN). pp. 1–8 (June 2023). https://doi.org/10.1109/IJCNN54540.2023.10191584

work page doi:10.1109/ijcnn54540.2023.10191584 2023
[31]

In: International Joint Conference on Neural Networks (IJCNN)

Laroca, R., Severo, E., Zanlorensi, L.A., Oliveira, L.S., Gonçalves, G.R., Schwartz, W.R., Menotti, D.: A robust real-time automatic license plate recognition based on the YOLO detector. In: International Joint Conference on Neural Networks (IJCNN). pp. 1–10 (July 2018).https://doi.org/10.1109/IJCNN.2018.8489629

work page doi:10.1109/ijcnn.2018.8489629 2018
[32]

In: Iberoamerican Congress on Pattern Recognition (CIARP)

Laroca, R., Zanlorensi, L.A., Estevam, V., Minetto, R., Menotti, D.: Leveraging model fusion for improved license plate recognition. In: Iberoamerican Congress on Pattern Recognition (CIARP). pp. 60–75 (Nov 2023).https://doi.org/10.1007/ 978-3-031-49249-5_5

2023
[33]

A.; Gonçalves, G

Laroca, R., Zanlorensi, L.A., Gonçalves, G.R., Todt, E., Schwartz, W.R., Menotti, D.: An efficient and layout-independent automatic license plate recognition system based on the YOLO detector. IET Intelligent Transport Systems15(4), 483–503 (2021).https://doi.org/10.1049/itr2.12030

work page doi:10.1049/itr2.12030 2021
[34]

https://doi.org/10.1049/itr2.70086

Laroca, R., Estevam, V., Moreira, G.J.P., Minetto, R., Menotti, D.: Advancing multinational license plate recognition through synthetic and real data fusion: A comprehensiveevaluation.IETIntelligentTransportSystems19(1),e70086(2025). https://doi.org/10.1049/itr2.70086

work page doi:10.1049/itr2.70086 2025
[35]

Journal of the Brazilian Computer Society32(1), 783–799 (2026)

Lima, G.E., Nascimento, V., Santos, E., Nascimento Jr., E., Laroca, R., Menotti, D.: Toward unified fine-grained vehicle classification and automatic license plate recognition. Journal of the Brazilian Computer Society32(1), 783–799 (2026). https://doi.org/10.5753/jbcs.2026.5899

work page doi:10.5753/jbcs.2026.5899 2026
[36]

In: International Conference on Multimedia Modeling

Liu,Y.Y.,Liu,Q.,Chen,F.,Yin,X.C.:Irregularlicenseplaterecognitionviaglobal information integration. In: International Conference on Multimedia Modeling. pp. 325–339 (2024).https://doi.org/10.1007/978-3-031-53308-2_24

work page doi:10.1007/978-3-031-53308-2_24 2024
[37]

In: International Conference on Learning Representations (ICLR)

Loshchilov, I., Hutter, F.: SGDR: Stochastic gradient descent with warm restarts. In: International Conference on Learning Representations (ICLR). pp. 1–16 (2017)

2017
[38]

In: International Conference on Learning Representations (ICLR)

Loshchilov, I., Hutter, F.: Decoupled weight decay regularization. In: International Conference on Learning Representations (ICLR). pp. 1–19 (2019)

2019
[39]

In: IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)

Maier, A., Moussa, D., Spruck, A., Seiler, J., Riess, C.: Reliability scoring for the recognition of degraded license plates. In: IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS). pp. 1–8 (2022).https: //doi.org/10.1109/AVSS56176.2022.9959390 ICPR 2026 Competition on Low-Resolution License Plate Recognition 19

work page doi:10.1109/avss56176.2022.9959390 2022
[40]

In: International Conference on Learning Representations (ICLR)

Micikevicius, P., et al.: Mixed precision training. In: International Conference on Learning Representations (ICLR). pp. 1–12 (2018)

2018
[41]

In: IEEE International Conference on Image Processing (ICIP)

Moussa, D., Maier, A., Spruck, A., Seiler, J., Riess, C.: Forensic license plate recog- nition with compression-informed transformers. In: IEEE International Conference on Image Processing (ICIP). pp. 406–410 (Oct 2022).https://doi.org/10.1109/ ICIP46576.2022.9897178

work page arXiv 2022
[42]

Computer Vision and Image Understanding256, 104361 (2025).https://doi.org/10.1016/ j.cviu.2025.104361

Na, K., Oh, J., Cho, Y., Kim, B., Cho, S., Choi, J., Kim, I.: MF-LPR2: Multi- frame license plate image restoration and recognition using optical flow. Computer Vision and Image Understanding256, 104361 (2025).https://doi.org/10.1016/ j.cviu.2025.104361

work page arXiv 2025
[43]

In: Con- ference on Graphics, Patterns and Images (SIBGRAPI)

Nascimento, V., Laroca, R., Lambert, J.A., Schwartz, W.R., Menotti, D.: Combin- ing attention module and pixel shuffle for license plate super-resolution. In: Con- ference on Graphics, Patterns and Images (SIBGRAPI). pp. 228–233 (Oct 2022). https://doi.org/10.1109/SIBGRAPI55357.2022.9991753

work page doi:10.1109/sibgrapi55357.2022.9991753 2022
[44]

O., Schwartz, W

Nascimento, V., Laroca, R., Ribeiro, R.O., Schwartz, W.R., Menotti, D.: En- hancing license plate super-resolution: A layout-aware and character-driven ap- proach. Conference on Graphics, Patterns and Images (SIBGRAPI) pp. 1–6 (2024). https://doi.org/10.1109/SIBGRAPI62404.2024.10716303

work page doi:10.1109/sibgrapi62404.2024.10716303 2024
[45]

Journal of the Brazilian Computer Society1(31), 435– 449 (2025).https://doi.org/10.5753/jbcs.2025.5159

Nascimento, V., Lima, G.E., Ribeiro, R.O., Schwartz, W.R., Laroca, R., Menotti, D.: Toward advancing license plate super-resolution in real-world scenarios: A dataset and benchmark. Journal of the Brazilian Computer Society1(31), 435– 449 (2025).https://doi.org/10.5753/jbcs.2025.5159

work page doi:10.5753/jbcs.2025.5159 2025
[46]

IEEE Access9, 101065–101077 (2021).https://doi.org/10

Oliveira, I.O., Laroca, R., Menotti, D., Fonseca, K.V.O., Minetto, R.: Vehicle-Rear: Anewdatasettoexplorefeaturefusionforvehicleidentificationusingconvolutional neural networks. IEEE Access9, 101065–101077 (2021).https://doi.org/10. 1109/ACCESS.2021.3097964

work page arXiv 2021
[47]

arXiv preprint arXiv:2110.11314 (2021)

Olpadkar, K., Gavas, E.: Center loss regularization for continual learning. arXiv preprint arXiv:2110.11314 (2021)

work page arXiv 2021
[48]

OpenALPR: OpenALPR-BR dataset.https://github.com/openalpr/ benchmarks/tree/master/endtoend/br(2016)

2016
[49]

Transactions on Machine Learning Research (2024),https://openreview.net/ forum?id=a68SUt6zFt

Oquab, M., et al.: DINOv2: Learning robust visual features without supervision. Transactions on Machine Learning Research (2024),https://openreview.net/ forum?id=a68SUt6zFt

2024
[50]

Neurocomputing580, 127426 (2024)

Pan, Y., Tang, J., Tjahjadi, T.: LPSRGAN: Generative adversarial networks for super-resolution of license plate image. Neurocomputing580, 127426 (2024). https://doi.org/10.1016/j.neucom.2024.127426

work page doi:10.1016/j.neucom.2024.127426 2024
[51]

Ex- pert Systems with Applications243, 122878 (2024).https://doi.org/10.1016/ j.eswa.2023.122878

Rao, Z., Yang, D., Chen, N., Liu, J.: License plate recognition system in un- constrained scenes via a new image correction scheme and improved CRNN. Ex- pert Systems with Applications243, 122878 (2024).https://doi.org/10.1016/ j.eswa.2023.122878

work page arXiv 2024
[52]

In: Medical Image Compu ting and Computer-Assisted Intervention – MICCAI 2015

Ronneberger, O., Fischer, P., Brox, T.: U-Net: Convolutional networks for biomed- ical image segmentation. In: International Conference on Medical Image Comput- ing and Computer Assisted Intervention (MICCAI). pp. 234–241 (2015).https: //doi.org/10.1007/978-3-319-24574-4_28

work page doi:10.1007/978-3-319-24574-4_28 2015
[53]

IEEE Transactions on In- telligent Transportation Systems24(9), 9203–9216 (2023).https://doi.org/10

Schirrmacher, F., Lorch, B., Maier, A., Riess, C.: Benchmarking probabilistic deep learning methods for license plate recognition. IEEE Transactions on In- telligent Transportation Systems24(9), 9203–9216 (2023).https://doi.org/10. 1109/TITS.2023.3278533

work page arXiv 2023
[54]

IEEE Transac- 20 R

Shi, B., Bai, X., Yao, C.: An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition. IEEE Transac- 20 R. Laroca et al. tions on Pattern Analysis and Machine Intelligence39(11), 2298–2304 (Nov 2017). https://doi.org/10.1109/TPAMI.2016.2646371

work page doi:10.1109/tpami.2016.2646371 2017
[55]

In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

Shrivastava, A., Gupta, A., Girshick, R.: Training region-based object detectors with online hard example mining. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 761–769 (2016).https://doi.org/10.1109/ CVPR.2016.89

2016
[56]

IEEE Transactions on Intelligent Transportation Systems23(6), 5693–5703 (2022).https://doi.org/10.1109/TITS.2021.3055946

Silva, S.M., Jung, C.R.: A flexible approach for automatic license plate recogni- tion in unconstrained scenarios. IEEE Transactions on Intelligent Transportation Systems23(6), 5693–5703 (2022).https://doi.org/10.1109/TITS.2021.3055946

work page doi:10.1109/tits.2021.3055946 2022
[57]

DINOv3

Siméoni, O., et al.: DINOv3. arXiv preprint arXiv:2508.10104 pp. 1–67 (2025)

work page internal anchor Pith review arXiv 2025
[58]

In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

Sun, K., Xiao, B., Liu, D., Wang, J.: Deep high-resolution representation learning for human pose estimation. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 5686–5696 (2019).https://doi.org/10.1109/ CVPR.2019.00584

work page arXiv 2019
[59]

Machine Learning and Knowledge Extraction5(4), 1680–1716 (2023)

Terven, J., Córdova-Esparza, D.M., Romero-González, J.A.: A comprehensive re- view of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and YOLO-NAS. Machine Learning and Knowledge Extraction5(4), 1680–1716 (2023)

2023
[60]

Ultralytics: YOLOv11 (2026),https://docs.ultralytics.com/models/yolo11/, accessed: 2026-03-30

2026
[61]

In: International Conference on Neural Information Processing Systems (NeurIPS)

Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, L., Polosukhin, I.: Attention is all you need. In: International Conference on Neural Information Processing Systems (NeurIPS). p. 6000–6010 (2017)

2017
[62]

Journal of Visual Communication and Image Represen- tation104, 104314 (2024).https://doi.org/10.1016/j.jvcir.2024.104314

Wei, C., Han, F., Fan, Z., Shi, L., Peng, C.: Efficient license plate recognition in unconstrained scenarios. Journal of Visual Communication and Image Represen- tation104, 104314 (2024).https://doi.org/10.1016/j.jvcir.2024.104314

work page doi:10.1016/j.jvcir.2024.104314 2024
[63]

IEEE Access8, 91661–91675 (2020)

Weihong, W., Jiaoyang, T.: Research on license plate recognition algorithms based on deep learning in complex environment. IEEE Access8, 91661–91675 (2020). https://doi.org/10.1109/ACCESS.2020.2994287

work page doi:10.1109/access.2020.2994287 2020
[64]

International Joint Conference on Neural Networks (IJCNN) pp

Wojcik, L., Machoski, E.A.F., Nascimento Jr., E., Laroca, R., Menotti, D.: LPLCv2: An expanded dataset for fine-grained license plate legibility classifica- tion. International Joint Conference on Neural Networks (IJCNN) pp. 1–7 (2026)

2026
[65]

In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

Woo, S., Debnath, S., Hu, R., Chen, X., Liu, Z., Kweon, I.S., Xie, S.: ConvNeXt V2: Co-designing and scaling convnets with masked autoencoders. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 16133– 16142 (2023).https://doi.org/10.1109/CVPR52729.2023.01548

work page doi:10.1109/cvpr52729.2023.01548 2023
[66]

Patterns3(7), 100543 (2022).https://doi.org/10.1016/j.patter.2022.100543

Xu, Z., Escalera, S., Pavão, A., Richard, M., Tu, W.W., Yao, Q., Zhao, H., Guyon, I.: Codabench: Flexible, easy-to-use, and reproducible meta-benchmark platform. Patterns3(7), 100543 (2022).https://doi.org/10.1016/j.patter.2022.100543

work page doi:10.1016/j.patter.2022.100543 2022
[67]

In: IEEE/CVF International Conference on Computer Vision (ICCV)

Ye,X.,Du,Y.,Tao,Y.,Chen,Z.:TextSSR:Diffusion-baseddatasynthesisforscene text recognition. In: IEEE/CVF International Conference on Computer Vision (ICCV). pp. 17464–17473 (2025)

2025
[68]

In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2026)

Ye, X., Du, Y., Zhang, J., Li, C., LYU, J., Chen, Z.: What’s wrong with synthetic data for scene text recognition? a strong synthetic engine with diverse simulations and self-evolution. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2026)

2026
[69]

arXiv preprint arXiv:2208.11247 pp

Zhang, D., Huang, F., Liu, S., Wang, X., Jin, Z.: SwinFIR: Revisiting the SwinIR with fast fourier convolution and improved training for image super-resolution. arXiv preprint arXiv:2208.11247 pp. 1–14 (2023)

work page arXiv 2023
[70]

In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

Zhu, X., Hu, H., Lin, S., Dai, J.: Deformable convnets v2: More deformable, better results. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 9300–9308 (2019).https://doi.org/10.1109/CVPR.2019.00953

work page doi:10.1109/cvpr.2019.00953 2019