Work

A Practical Cost-Benefit Analysis of Pre-trained Model Selection

Year: 2025

Type: article

Abstract: This paper provides a systematic framework for selecting pre-trained computer vision models in enterprise applications by evaluating YOLO, MobileNet, ResNet, and EfficientNet across accuracy, speed, and cost dimensions. Through comparative analysis, we demonstrate cost variations up to 300% between models with similar accuracy, and provide actionable recommendations for cloud versus edge deployment strategies. A retail case study shows 90%+ cost savings through optimized model selection.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.17579325

Author: Amer, Mahmoud

Language: English

Topic: Advanced Neural Network Applications

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W7105173525

FWCI:

Cites:

Cited by:

Related to: