ANALISIS PERBANDINGAN YOLOv8 DAN EFFICIENTDET UNTUK DETEKSI OBJEK BERGERAK DI VIDEO SURVEILLANCE

Luh Putu Ary Sri Tjahyanti, Made Santo Gitakarma

Abstract


Sistem pengawasan video memerlukan deteksi objek bergerak yang cepat dan akurat, namun model deep learning memiliki karakteristik kinerja yang berbeda-beda. Penelitian ini bertujuan membandingkan performa YOLOv8 dan EfficientDet untuk deteksi objek bergerak di video surveillance. Metode yang digunakan meliputi preprocessing 5000 frame video dari dataset UA-DETRAC dan VisDrone, pelabelan objek (mobil, sepeda motor, dan pejalan kaki), serta pelatihan model dengan batch size 16 selama 100 epoch. Hasil utama menunjukkan YOLOv8n mencapai kecepatan inferensi 78 FPS dengan mAP 0.72, sedangkan EfficientDet-D3 mencapai 42 FPS dengan mAP 0.81. Pada deteksi objek berukuran kecil (<32×32 piksel), EfficientDet unggul dengan recall 0.68 dibandingkan YOLOv8 sebesar 0.51. Kontribusi utama penelitian ini adalah: (1) menyediakan perbandingan sistematis pertama antara YOLOv8 dan EfficientDet khusus untuk deteksi objek bergerak pada video surveillance; (2) memberikan tolok ukur kuantitatif (FPS, mAP, recall untuk objek kecil) sebagai referensi praktis bagi pengembang sistem pengawasan; serta (3) merumuskan panduan pemilihan model berdasarkan prioritas real-time atau presisi. Kesimpulannya, YOLOv8 direkomendasikan untuk aplikasi real-time (≥60 FPS), sementara EfficientDet cocok untuk skenario yang memprioritaskan presisi tinggi.

Full Text:

PDF

References


APPLIED SCIENCES. (2026). Systematic evaluation of YOLOv8 variants for UAV-based object detection. Applied Sciences, *16*(7), 3559.

CHAKRABORTY, A., et al. (2025). Profiling concurrent vision inference workloads on NVIDIA Jetson -- Extended. arXiv.

COCO EVALUATOR. (2025). COCO object detection evaluation metrics. COCO Consortium.

ELVIANI, U., HANAVI, H., & HIDAYAT, F. (2025). Lost centroid ID in CCTV analytics for slow moving object. COELITE, *4*(2). https://doi.org/10.17509/coelite.v4i2.85752

ICHORA. (2025). Real-time object detection: A comparative analysis of YOLO, SSD, and EfficientDet algorithms. In *7th International Congress on Human-Computer Interaction, Optimization and Robotic Applications (ICHORA)*. Ankara, Turkey. https://doi.org/10.1109/ichora65333.2025.11017287

IEEE. (2025). Intelligent video surveillance systems with violence detection. IEEE Access. https://ieeexplore.ieee.org/document/11177151

JOCHER, G., CHAURASIA, A., & QIU, J. (2023). Ultralytics YOLOv8. Ultralytics. https://github.com/ultralytics/ultralytics

KHAIRWA, A., & THANGAVELU, A. (2024). Moving object detection for surveillance video frames using two stage multi-scale residual convolution neural networks. International Journal of Grid and Utility Computing, *15*(3/4), 253-262.

KHAIRWA, A., & THANGAVELU, S. (2024). GPU-based deep learning optimization for real-time object detection systems. Journal of Artificial Intelligence and Computing.

LIU, Y., GENG, L., ZHANG, W., GONG, Y., & XU, Z. (2021). Survey of video based small target detection. Journal of Image and Graphics, *9*(4), 122-134.

LIU, Y., et al. (2024). Research on the perception method of tiny objects in low-light and wide-field video. Scientific Reports, *14*, 17249.

LV, C., MITTAL, U., MADAAN, V., & AGRAWAL, P. (2024). Vehicle detection and classification using an ensemble of EfficientDet and YOLOv8. PeerJ Computer Science, *10*, e2233. https://doi.org/10.7717/peerj-cs.2233

NVIDIA. (2024). Nsight systems user guide. NVIDIA Developer Documentation.

ODEI. (2025). ODEI: Object detector efficiency index. AI, *6*(7), 141. https://doi.org/10.3390/ai6070141

PADILLA, R., PASSOS, W. L., DIAS, T. L. B., NETTO, S. L., & DA SILVA, E. A. B. (2021). A comparative analysis of object detection metrics with a companion open-source toolkit. Electronics, *10*(3), 279.

SABELNIKOV, P. (2025). Automated detection of foreign objects in video scenes with dynamic obstacles. Science and Innovation, *21*(5), 62-75. https://doi.org/10.15407/scine21.05.062

TAN, M., PANG, R., & LE, Q. V. (2020). EfficientDet: Scalable and efficient object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 10781-10790).

ULTRALYTICS. (2025). EfficientDet vs YOLOv8: A comprehensive technical comparison of object detection architectures. Ultralytics Documentation.

ZHAO, Z. Q., ZHENG, P., XU, S. T., & WU, X. (2021). Object detection with deep learning: A review. IEEE Transactions on Neural Networks and Learning Systems.




DOI: https://doi.org/10.37637/komteks.v5i1.3003

Refbacks

  • There are currently no refbacks.


Published By:

PRODI TEKNOLOGI INFORMASI
FAKULTAS TEKNIK
UNIVERSITAS PANJI SAKTI

Creative Commons License

Komteks is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.