Zobrazit v EDS

Parallel Processing Optimization for VITON-HD Virtual Try-On: Performance Analysis and Resource Management Framework.

Uloženo v:
Podrobná bibliografie
Název: Parallel Processing Optimization for VITON-HD Virtual Try-On: Performance Analysis and Resource Management Framework.
Autoři: P., Thabitha, S., Ramalakshmi, Keerthika Velusamy
Zdroj: Grenze International Journal of Engineering & Technology (GIJET); Jan2026, Vol. 12 Issue Part2, p2719-2725, 7p
Témata: PARALLEL processing, RESOURCE management, SCALABILITY, COMPUTER performance, PARALLEL programming, ELECTRONIC commerce
Abstrakt: Virtual try-on (VTO) systems face significant computational bottlenecks due to high processing latency and limited scalability, hindering real-time e-commerce applications. This paper presents a CPU-based parallel processing framework that transforms the sequential VITON-HD inference pipeline into concurrent processes using process isolation techniques. Our approach employs a hybrid execution strategy with ProcessPoolExecutor for isolated parallel processes and ThreadPoolExecutor as a fallback mechanism, ensuring system reliability across different deployment environments. The framework incorporates mathematical performance modeling using Amdahl's Law to predict theoretical speedups and guide resource allocation decisions. We implement resource-aware worker scheduling that dynamically adjusts based on available system resources to prevent overload while maximizing throughput. Experimental results on a 12-core system using 2 parallel workers demonstrate a 1.04× speedup with 52.1% parallel efficiency, reducing processing time by 6.7 seconds for batch operations. While the observed performance gains are below theoretical predictions due to process creation overhead and I/O limitations, the framework achieves 100% success rate in batch processing and provides a practical solution for scalable virtual try-on systems without requiring expensive GPU infrastructure. The system offers significant improvements in CPU utilization (78% vs 65%) and enables concurrent processing of multiple person-garment combinations, making VTO technology more accessible for commercial deployment. [ABSTRACT FROM AUTHOR]
Copyright of Grenze International Journal of Engineering & Technology (GIJET) is the property of GRENZE Scientific Society and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Databáze: Complementary Index
Popis
Abstrakt:Virtual try-on (VTO) systems face significant computational bottlenecks due to high processing latency and limited scalability, hindering real-time e-commerce applications. This paper presents a CPU-based parallel processing framework that transforms the sequential VITON-HD inference pipeline into concurrent processes using process isolation techniques. Our approach employs a hybrid execution strategy with ProcessPoolExecutor for isolated parallel processes and ThreadPoolExecutor as a fallback mechanism, ensuring system reliability across different deployment environments. The framework incorporates mathematical performance modeling using Amdahl's Law to predict theoretical speedups and guide resource allocation decisions. We implement resource-aware worker scheduling that dynamically adjusts based on available system resources to prevent overload while maximizing throughput. Experimental results on a 12-core system using 2 parallel workers demonstrate a 1.04× speedup with 52.1% parallel efficiency, reducing processing time by 6.7 seconds for batch operations. While the observed performance gains are below theoretical predictions due to process creation overhead and I/O limitations, the framework achieves 100% success rate in batch processing and provides a practical solution for scalable virtual try-on systems without requiring expensive GPU infrastructure. The system offers significant improvements in CPU utilization (78% vs 65%) and enables concurrent processing of multiple person-garment combinations, making VTO technology more accessible for commercial deployment. [ABSTRACT FROM AUTHOR]
ISSN:23955287