Comparative Performance Analysis of YOLOv12 and RF-DETR in Face Detection
DOI:
https://doi.org/10.63158/journalisi.v8i2.1561Keywords:
YOLOv12, RF-DETR, WIDER FACE, Inference Latency, Edge DeploymentAbstract
Face detection in dense and occluded environments remains a significant challenge in computer vision. This study compares the CNN-based YOLOv12 and the Transformer-based RF-DETR to determine the optimal balance between accuracy and latency for resource-constrained edge computing. Using the WIDER FACE dataset and an NVIDIA T4 GPU, multiple model variants were evaluated. Due to GPU memory constraints during training of the RF-DETR Medium variant, a standardized batch size of 8 was implemented across all models. To ensure methodological rigor, quantitative metrics (precision, recall, F1-score, mAP) were strictly assessed on the validation set. Concurrently, a 100-image subset of the test set was used exclusively for inference efficiency benchmarking, completely separate from detection evaluation. Results indicate YOLOv12X achieved superior overall detection performance (F1-score: 0.764, mAP@50:95: 0.440), significantly outperforming RF-DETR Medium. For real-time applications, YOLOv12M demonstrated the highest efficiency (36.17 FPS vs. 23.32 FPS). Qualitatively, YOLOv12 maintained high sensitivity in crowded scenes, whereas RF-DETR provided stable small-scale face detection despite its lower recall. Overall, under these constrained-hardware conditions, YOLOv12 appears to be a highly viable solution for surveillance systems, while RF-DETR offers a stable alternative for small-object detection when computational overhead and training budgets are less restrictive.
Downloads
References
[1] A. S. Sanchez-Moreno, J. Olivares-Mercado, A. Hernandez-Suarez, K. Toscano-Medina, G. Sanchez-Perez, and G. Benitez-Garcia, “Efficient face recognition system for operating in unconstrained environments,” J. Imaging, vol. 7, no. 9, p. 161, Sep. 2021, doi: 10.3390/jimaging7090161.
[2] Z. Yu, H. Huang, W. Chen, Y. Su, Y. Liu, and X. Wang, “YOLO-FaceV2: A scale and occlusion aware face detector,” Pattern Recognit., vol. 155, p. 110714, Nov. 2024, doi: 10.1016/j.patcog.2024.110714.
[3] B. Balachander, B. Sarveswari, S. N. S, C. S. Sowmiya, and R. Rajeshwari, “Performance analysis of lightweight YOLOv12 framework for object detection using real time Web camera based inputs,” in 2025 International Conference on Recent Innovation in Science Engineering and Technology (ICRISET), Aug. 2025, pp. 1–6. doi: 10.1109/ICRISET64803.2025.11252243.
[4] D. Mamieva, A. B. Abdusalomov, M. Mukhiddinov, and T. K. Whangbo, “Improved face detection method via learning small faces on hard images based on a deep learning approach,” Sensors, vol. 23, no. 1, p. 502, Jan. 2023, doi: 10.3390/s23010502.
[5] H. Du, H. Shi, D. Zeng, X.-P. Zhang, and T. Mei, “The elements of end-to-end deep face recognition: A survey of recent advances,” ACM Comput Surv, vol. 54, no. 10s, p. 212, Sep. 2022, doi: 10.1145/3507902.
[6] Q. Xu, Z. Zhu, H. Ge, Z. Zhang, and X. Zang, “Effective face detector based on YOLOv5 and superresolution reconstruction,” Comput. Math. Methods Med., vol. 2021, no. 1, p. 7748350, 2021, doi: 10.1155/2021/7748350.
[7] M. E. Atik and M. Arkali, “Benchmarking YOLO and Transformer-based detectors for olive tree crown identification in UAV imagery,” Geomatics, vol. 6, no. 2, p. 22, Feb. 2026, doi: 10.3390/geomatics6020022.
[8] N. Dahiya et al., “Optimised RFO tuned RF-DETR model for precision urine microscopy for renal and systemic disease diagnosis,” Sci. Rep., vol. 15, no. 1, p. 25842, Jul. 2025, doi: 10.1038/s41598-025-11725-0.
[9] M. L. Ali and Z. Zhang, “The YOLO framework: A comprehensive review of evolution, applications, and benchmarks in object detection,” Computers, vol. 13, no. 12, p. 336, Dec. 2024, doi: 10.3390/computers13120336.
[10] R. Sapkota, Z. Meng, M. Churuvija, X. Du, Z. Ma, and M. Karkee, “Comprehensive performance evaluation of YOLOv12, YOLO11, YOLOv10, YOLOv9 and YOLOv8 on detecting and counting fruitlet in complex orchard environments,” Agric. Commun., vol. 4, no. 1, p. 100125, Mar. 2026, doi: 10.1016/j.agrcom.2026.100125.
[11] Y. Sun, Z. Sun, and W. Chen, “The evolution of object detection methods,” Eng. Appl. Artif. Intell., vol. 133, p. 108458, Jul. 2024, doi: 10.1016/j.engappai.2024.108458.
[12] K. Huang, M. Wen, C. Wang, and L. Ling, “T-SSD: A transformer-based single-stage multi-scale sampling object detector,” in Proceedings of 2023 the 13th International Workshop on Computer Science and Engineering, WCSE, 2023. doi: 10.18178/wcse.2023.06.022.
[13] M. Chaman, A. E. Maliki, H. Dahou, and A. Hadjoudja, “Benchmarking YOLO-based deep learning models for real-time object detection in hybrid ADAS and intelligent transportation systems,” Results Eng., vol. 29, p. 108942, Mar. 2026, doi: 10.1016/j.rineng.2025.108942.
[14] J. M. Villarroel, L. C. De Jesus, J. Ancheta, H. Villaruel, and L. Samaniego, “Comparative analysis of YOLOv12 and RF-DETR models for coffee leaf disease detection using Roboflow,” in 2025 IEEE 14th Global Conference on Consumer Electronics (GCCE), Sep. 2025, pp. 1493–1494. doi: 10.1109/GCCE65946.2025.11274815.
[15] M. S. Aqilla, M. G. Abdurahman, A. B. Hendry, M. R. Arjasubrata, and M. D. Sulistiyo, “A comparative analysis of YOLOv12 and RFDETR for weapon detection in CCTV footage,” in 2025 5th International Conference of Science and Information Technology in Smart Administration (ICSINTESA), Nov. 2025, pp. 35–40. doi: 10.1109/ICSINTESA68165.2025.11413754.
[16] S. Zanniko, J. Cahyo, A. A. S. Gunawan, and R. C. Pradana, “Comparative analysis of RF-DETR and YOLOv12 in breast cancer detection and classification,” in 2025 International Conference on Information Management and Technology (ICIMTech), Aug. 2025, pp. 246–251. doi: 10.1109/ICIMTech67074.2025.11265111.
[17] T. Sabrina, I. Damayanti, I. Mahardika, M. R. Arjasubrata, and M. D. Sulistiyo, “Comparative analysis of CNN and transformer models for cigarette and e-cigarette detection,” in 2025 8th International Seminar on Research of Information Technology and Intelligent Systems (ISRITI), Dec. 2025, pp. 345–350. doi: 10.1109/ISRITI68345.2025.11393399.
[18] C. L. Buana, G. F. Shidik, M. N. Ubaidillah, Y. R. Grafer, F. A. Kristiyan, and E. J. Kusuma, “Comparison of YOLO and RF-DETR performance in detecting personal protective equipment in construction environments using tokenization,” in 2025 International Seminar on Application for Technology of Information and Communication (iSemantic), Sep. 2025, pp. 151–157. doi: 10.1109/ISemantic67418.2025.11292414.
[19] U. Aymon, N. S. Kamarudin, and A. F. Ab. Nasir, “Facial expression recognition with YOLOv11 and YOLOv12: A comparative study,” in 2025 IEEE 9th International Conference on Software Engineering & Computer Systems (ICSECS), Oct. 2025, pp. 18–23. doi: 10.1109/ICSECS65227.2025.11279248.
[20] Q. Tang, Y. Li, Y. Cai, X. Peng, and X. Liu, “Face detection based on DF-Net,” Electronics, vol. 12, no. 19, p. 4021, Sep. 2023, doi: 10.3390/electronics12194021.
[21] M. G. Ragab et al., “A comprehensive systematic review of YOLO for medical object detection (2018 to 2023),” IEEE Access, vol. 12, pp. 57815–57836, 2024, doi: 10.1109/ACCESS.2024.3386826.
[22] Y. Ji et al., “Transmission line defect detection algorithm based on improved YOLOv12,” Electronics, vol. 14, no. 12, p. 2432, Jun. 2025, doi: 10.3390/electronics14122432.
[23] L. T. Ramos and A. D. Sappa, “A comprehensive analysis of YOLO architectures for tomato leaf disease identification,” Sci. Rep., vol. 15, no. 1, p. 26890, Jul. 2025, doi: 10.1038/s41598-025-11064-0.
[24] N. Ghosh and G. Mandal, “Classification of canine dermatological diseases using YOLOv12 with R-ELAN,” in 2025 International Conference on Computational Intelligence and Knowledge Economy (ICCIKE), Nov. 2025, pp. 542–546. doi: 10.1109/ICCIKE67021.2025.11318252.
[25] A. S. Silva, F. H. A. Moraes Neto, P. H. Ferreira, and D. B. Costa, “CNN-based YOLOv12 for damage assessment in residential roofs,” IEEE Access, vol. 13, pp. 193311–193322, 2025, doi: 10.1109/ACCESS.2025.3629630.
[26] N. Deluxni, P. Sudhakaran, M. Alsafyani, and A. Yousef, “Underwater debris segmentation using improved YOLOv12s with recursive efficient layer aggregation and FlashAttention for autonomous underwater vehicle,” IEEE Access, vol. 13, pp. 200239–200252, 2025, doi: 10.1109/ACCESS.2025.3636283.
[27] A. A. Murat and M. S. Kiran, “A comprehensive review on YOLO versions for object detection,” Eng. Sci. Technol. Int. J., vol. 70, p. 102161, Oct. 2025, doi: 10.1016/j.jestch.2025.102161.
[28] J. Terven, D.-M. Córdova-Esparza, and J.-A. Romero-González, “A comprehensive review of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and YOLO-NAS,” Mach. Learn. Knowl. Extr., vol. 5, no. 4, pp. 1680–1716, Nov. 2023, doi: 10.3390/make5040083.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Journal of Information Systems and Informatics
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors Declaration
- The Authors certify that they have read, understood, and agreed to the Journal of Information Systems and Informatics (JournalISI) submission guidelines, policies, and submission declaration. The submission has been prepared using the provided template.
- The Authors certify that all authors have approved the publication of this manuscript and that there is no conflict of interest.
- The Authors confirm that the manuscript is their original work, has not received prior publication, is not under consideration for publication elsewhere, and has not been previously published.
- The Authors confirm that all authors listed on the title page have contributed significantly to the work, have read the manuscript, attest to the validity and legitimacy of the data and its interpretation, and agree to its submission.
- The Authors confirm that the manuscript is not copied from or plagiarized from any other published work.
- The Authors declare that the manuscript will not be submitted for publication in any other journal or magazine until a decision is made by the journal editors.
- If the manuscript is finally accepted for publication, the Authors confirm that they will either proceed with publication immediately or withdraw the manuscript in accordance with the journal’s withdrawal policies.
- The Authors agree that, upon publication of the manuscript in this journal, they transfer copyright or assign exclusive rights to the publisher, including commercial rights
