Energy-Accuracy Tradeoff in Energy-Aware YOLOv11 Optimization for Edge AI

View at publisher Summarise with VIRAT
Authors: Hariharan P Hariharan P, Anju S. Pillai
Cited by: 0
References: 13

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Nowadays, running a deep learning models in environments with limited resources, like edge devices and servers, has become difficult due to its power and hardware restrictions. It has become important as edge computation and embedded AI technology increase at a rapid pace. Edge devices running for AI application generally operates on tight power budgets in returns needed to have a careful balance between energy use and model accuracy while also maintaining strict limits on power, computation, and inference time. In the work below, the energy-accuracy tradeoffs YOLOv11 models for edge deployment are thoroughly examined. When optimisation strategies like quantization, pruning and hybrid techniques. When applied apart from minimal degradation, findings showed$2 x$of energy savings compared to baseline model with retaining 95% and 99% of precision and recall. Apart from this results, hybrid techniques shown 10x speed up compared to baseline model. This new result provides as with a sustainable method for edge deployment and also shows how structure aware pruning is needed while pruning related optimisation is applied.

Fields of Study & Subjects

Advanced Neural Network Applications IoT and Edge/Fog Computing Energy Efficiency in Computing

Publication details

Journal / source
International Conference on Intelligent Systems for Communication, IoT and Security (ICISCoIS)
Pages
1–5
Publication date
8 January 2026
Language
EN