Analisis Perbandingan Metode RF-DETR dan YOLO26 Dalam Mendeteksi Objek Jalan Berlubang: Comparative Analysis of RF-DETR and YOLO26 Methods in Detecting Pothole Road Objects

Fajar Hidayat; Parjito  Parjito

doi:10.57152/malcom.v6i4.2806

Authors

Fajar Hidayat Universitas Teknokrat Indonesia
Parjito Parjito Universitas Teknokrat Indonesia

DOI:

https://doi.org/10.57152/malcom.v6i4.2806

Keywords:

Deteksi Objek, Jalan Berlubang, RF-DETR, Visi Komputer, YOLO26

Abstract

Kerusakan jalan berupa lubang merupakan salah satu faktor utama yang dapat mengganggu keselamatan lalu lintas dan menurunkan kualitas infrastruktur. Deteksi manual memiliki keterbatasan dalam cakupan dan kecepatan, sehingga diperlukan solusi deteksi objek otomatis berbasis kecerdasan buatan. Penelitian ini membandingkan dua metode deteksi objek, yaitu YOLO26 dan RF-DETR, dalam mendeteksi objek jalan berlubang. Dataset yang digunakan terdiri dari 2771 dengan anotasi bounding box yang diperoleh dari Roboflow Universe. Dataset dibagi menjadi data, yaitu 70% data train, 20% data valid, dan 10% data test. Pada perbandingan pelatihan kedua model, YOLO26 lebih unggul dalam waktu pelatihan dibandingkan RF-DETR, namun RF-DETR lebih cepat mendapatkan konvergensi kurang dari 20 epoch. Pada kedua model dilakukan evaluasi menggunakan metrik standar seperti Precision, Recall, F1-Score, serta mean Average Precision (mAP@50 dan mAP@50-95). Hasil penelitian menunjukkan bahwa YOLO26 lebih ungul menghasilkan nilai precision tinggi, dengan YOLO26L mencapai precision sebesar 0,7398 dan mAP@50 sebesar 0,4846. Sedangkan RF-DETR unggul dalam recall dengan nilai 0,8089, menandakan kemampuan mendeteksi lebih banyak objek kecil, namun precision lebih rendah. Dengan demikian, YOLO26 lebih sesuai diterapkan pada sistem deteksi kerusakan jalan yang membutuhkan akurasi tinggi dan efisiensi komputasi, sementara RF-DETR tetap relevan untuk konteks yang menekankan cakupan deteksi luas.

Downloads

Download data is not yet available.

References

F. Yudaningrum, “IDENTIFIKASI JENIS KERUSAKAN JALAN (Studi Kasus Ruas Jalan Kedungmundu-Meteseh),” TEKNIKA, vol. XII, no. 2, pp. 16–23, Oct. 2017, doi: 10.26623/teknika.v12i2.638.

K. A. Sattar, M. Abdel-Nasser, S. El Ferik, and A. E. Taha, “Detecting Road Defects and Hazards in Metropolitan Environments Using Optimized Deep Learning Techniques,” in Transportation Research Procedia, Elsevier B.V., 2025, pp. 528–533. doi: 10.1016/j.trpro.2025.03.105.

S. Sa-ngiem, K. Dittakan, and S. Boonsiripant, “The road conditions detection using the convolutional neural network,” Indonesian Journal of Electrical Engineering and Computer Science, vol. 40, no. 1, pp. 327–345, Oct. 2025, doi: 10.11591/ijeecs.v40.i1.pp327-345.

M. Jakubec, E. Lieskovská, B. Bu?ko, and K. Zábovská, “Comparison of CNN-Based Models for Pothole Detection in Real-World Adverse Conditions: Overview and Evaluation,” Applied Sciences (Switzerland), vol. 13, no. 9, May 2023, doi: 10.3390/app13095810.

K. R. Ahmed, “Smart pothole detection using deep learning based on dilated convolution,” Sensors, vol. 21, no. 24, Dec. 2021, doi: 10.3390/s21248406.

M. Tokat and Z. Elmi, “Edge-Ready Pothole Detection: Accuracy–Throughput–Power Trade-offs Across CNN, Transformer, and Real-Time Detectors,” Mar. 2026, doi: 10.2139/ssrn.6353877.

. Wang, Y. Chen, Y. Gu, J. Liu, X. Zhu, and M. He, “The evolution of object detection from CNNs to transformers and multi-modal fusion,” Sci. Rep., vol. 16, no. 1, Dec. 2026, doi: 10.1038/s41598-026-37052-6.

A. B. Amjoud and M. Amrouch, “Object Detection Using Deep Learning, CNNs and Vision Transformers: A Review,” IEEE Access, vol. 11, pp. 35479–35516, 2023, doi: 10.1109/ACCESS.2023.3266093.

N. Bhavana, M. M. Kodabagi, B. M. Kumar, P. Ajay, N. Muthukumaran, and A. Ahilan, “POT-YOLO: Real-Time Road Potholes Detection Using Edge Segmentation-Based Yolo V8 Network,” IEEE Sens. J., vol. 24, no. 15, pp. 24802–24809, May 2024, doi: 10.1109/JSEN.2024.3399008.

L. V. Fortin and O. E. Llantos, “Performance Analysis of YOLO versions for Real-time Pothole Detection,” in Procedia Computer Science, Elsevier B.V., Mar. 2025, pp. 77–84. doi: 10.1016/j.procs.2025.03.013.

Y. A. Chairunisa and G. Kosala, “Performance Comparison of YOLOv9, YOLOv10, and YOLOv11 for Detecting Highway Potholes,” in 2025 International Conference on Data Science and Its Applications, ICoDSA 2025, Institute of Electrical and Electronics Engineers Inc., 2025, pp. 7–12. doi: 10.1109/ICoDSA67155.2025.11157220.

L. V. Fortin and O. E. Llantos, “Explainable AI for Pothole Detection: Comparing YOLOv9-tiny, YOLOv10-nano, and YOLOv11-nano through LayerCAM Visualization,” Institute of Electrical and Electronics Engineers (IEEE), Jan. 2026, pp. 557–561. doi: 10.1109/cyberscitech68397.2025.00081.

Y. Tian, Q. Ye, and D. Doermann, “YOLOv12: Attention-Centric Real-Time Object Detectors,” Feb. 2025, doi: 10.48550/arXiv.2502.12524.

R. Sapkota, R. H. Cheppally, A. Sharda, and M. Karkee, “YOLO26: Key Architectural Enhancements and Performance Benchmarking for Real-Time Object Detection,” Mar. 2026, doi: 10.48550/arXiv.2509.25164.

S. Chakrabarty, “YOLO26: An Analysis of NMS-Free End to End Framework for Real-Time Object Detection,” Mar. 2026, doi: 10.48550/arXiv.2601.12882.

Y. Zhao et al., “DETRs Beat YOLOs on Real-time Object Detection,” in Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, IEEE Computer Society, Jun. 2024, pp. 16965–16974. doi: 10.1109/CVPR52733.2024.01605.

I. Robinson, P. Robicheaux, M. Popov, D. Ramanan, and N. Peri, “RF-DETR: Neural Architecture Search for Real-Time Detection Transformers,” Feb. 2026, doi: 10.48550/arXiv.2511.09554.

S. P. Kurrey, M. Rai, and T. K. R. Bollu, “Intelligent Road Surface Assessment: A Comparative Deep Learning Approach using YOLOv12 and RF-DETR,” in International Conference on Communication Systems and Networks, COMSNETS, Institute of Electrical and Electronics Engineers Inc., Mar. 2026, pp. 745–750. doi: 10.1109/COMSNETS67989.2026.11418262.

R. Sapkota and M. Karkee, “Ultralytics YOLO Evolution: An Overview of YOLO26, YOLO11, YOLOv8 and YOLOv5 Object Detectors for Computer Vision and Pattern Recognition,” Mar. 2026, doi: 10.48550/arXiv.2510.09653.

J. Xiao, J. Chen, Y. Ning, and Y. Jiang, “Object Detection in Recent Years: An Overview,” in ACM International Conference Proceeding Series, Association for Computing Machinery, Jan. 2024, pp. 628–635. doi: 10.1145/3672758.3672862.

Roboflow, “Pothole Dataset.” Accessed: Apr. 18, 2026. [Online]. Available: https://universe.roboflow.com/potholes-r7qcn/pothole-jujbl/dataset/1.

R. Sapkota, R. H. Cheppally, A. Sharda, and M. Karkee, “RF-DETR Object Detection vs YOLOv12?: A Study of Transformer-based and CNN-based Architectures for Single-Class and Multi-Class Greenfruit Detection in Complex Orchard Environments Under Label Ambiguity,” Apr. 2025, doi: 10.48550/arXiv.2504.13099.

D. Raya Ismail, R. Rahmadewi Teknik Elektro, U. Singaperbangsa Karawang HSRonggo Waluyo, K. Telukjambe Tim, K. Karawang, and J. Barat, “SISTEM DETEKSI JALAN BERLUBANG SECARA REAL-TIME MENGGUNAKAN YOLOV11: INTEGRASI DATA DAN LOKASI MELALUI WEBSITE (STUDI KASUS?: DAERAH KARAWANG),” Jurnal Mahasiswa Teknik Informatika), vol. 9, no. 3, pp. 3953–3961, Jun. 2025, doi: 10.36040/jati.v9i3.13436.

J. Homepage et al., “MALCOM: Indonesian Journal of Machine Learning and Computer Science Implementation and Comparison of YOLO v8, v9, and v11 in Occupational Safety Regulations: Detecting Safety Helmet Usage in Construction Environments Implementasi Perbandingan YOLO v8, v9, dan v11 dalam Penerapan Tata Tertib K3: Deteksi Penggunaan Helm Keselamatan di Lingkungan Konstruksi,” vol. 6, no. 2, pp. 513–523, 2026, doi: 10.57152/malcom.v6i2.2554.

C. Zuo, N. Huang, C. Yuan, and Y. Li, “Pavement-DETR: A High-Precision Real-Time Detection Transformer for Pavement Defect Detection,” Sensors, vol. 25, no. 8, Apr. 2025, doi: 10.3390/s25082426.