Real-time computer-vision systems deployed on edge platforms are constrained as much by the surrounding softwarepipeline and the host–accelerator interface as by raw compute. This paper presents a reproducible, hardware-aware optimization studyof the lightweight YOLOv8n object detector on an NVIDIA Tesla T4 GPU and contrasts the measured GPU behavior against ananalytical projection of a Xilinx Zynq UltraScale+ (ZU9EG) FPGA accelerator. Applying TensorRT graph optimization—verticalConv–BatchNorm–activation layer fusion together with FP16 precision—raised single-stream inference throughput on the T4 from aPyTorch FP32 baseline of 62.92 FPS to 72.72 FPS, a 1.16× (15.6%) speed-up, while mean average precision on COCO val2017 fellonly marginally (mAP@0.5 from 52.8% to 52.6%). A controlled batch-size sweep then exposes a host-side preprocessing bottleneck:as the batch grows to 16, host CPU utilization saturates to 98%, while GPU utilization decreases to 8%, and end-to-end throughputdegrades to 2.05 effective FPS. We show analytically and empirically that this is a data-ingest (decode/resize) limitation of the hostpipeline rather than a compute limit of the accelerator. Finally, using device datasheets and published FPGA-accelerator figures, weproject that a spatial INT8 dataflow accelerator would deliver lower peak throughput (≈45 FPS) but markedly better energy efficiencyand deterministic per-frame latency, characteristics desirable for latency-critical edge deployment. All numerical claims are clearlyseparated into measured (GPU) and projected (FPGA) categories.
| Mualliflar | Tursunaliev, Ulugbek, Tulkinov, Bakhromjon |
|---|---|
| Jurnal | Innovation science and technologiy |
| Nashr sanasi | 2026-06-01 |
| Jild | 2 |
| Son | 6 |
| Til | Ingliz |
| DOI | 10.5281/zenodo.20781147 |
DOI: 10.5281/zenodo.20781147 · Maqolaning asl sahifasi
object detection; YOLOv8; TensorRT; layer fusion; FP16/INT8 quantization; GPU; FPGA; edge inference; energy efficiency; deterministic latency.
This study investigates the theoretical and practical aspects of developing and enhancing spatial information modelsfor sustainable regional development with consideration of environmental factors, using the Republic of…
This article analyzes the structural and factor decomposition of economic growth across the districts of Surxondaryoregion using the Solow–Swan dynamic growth model and the Cobb–Douglas production function. The study…
The relationship between university leadership styles and the development of an institutional quality assurance (QA)culture has become a central issue in higher education management. Although external accreditation…
This article examines the strategic directions for improving and transforming the tax administration system in theRepublic of Uzbekistan. The study analyzes issues related to further improving the quality of taxpayer…
This article analyzes the issues of increasing the efficiency of research and development work in the agricultural sectorof the Republic of Uzbekistan, the introduction of innovative developments into production, and…
This article provides a comprehensive scientific analysis of the fundamental and practical methodologies for identifying textsgenerated by Artificial Intelligence (AI) within academic and digital publishing ecosystems…
This article examines the sociolinguistic features of negative aesthetic evaluation units denoting physical appearancein English and Uzbek. The research focuses on lexical and phraseological units belonging to the…
This study examines cellular aerated concretes based on polymer compositions and their standardization as lightweightconstruction materials for the construction industry of Uzbekistan. The results of large-scale…
International remittances’ impact on Uzbekistan’s economic growth is examined using time-series analysis methods,including ADF, Johansen cointegration, VAR, and VECM, based on data for 2000–2025. The empirical findings…
In the context of increasingly complex cyber threats, there is an urgent need for innovative approaches capableof effectively identifying and mitigating potential risks. Neural networks offer a promising solution to…