GreenStableYolo: Optimizing Inference Time and Image Quality of Text-to-Image Generation

TL;DR

GreenStableYolo enhances text-to-image generation by optimizing parameters and prompts, significantly reducing inference time and increasing image quality through multi-objective optimization, thus advancing current state-of-the-art methods.

Contribution

It introduces GreenStableYolo, a novel approach that optimizes inference speed and image quality for Stable Diffusion using NSGA-II and Yolo, with a focus on balancing trade-offs.

Findings

266% reduction in GPU inference time

18% decrease in image quality compared to StableYolo

526% higher hypervolume indicating better optimization

Abstract

Tuning the parameters and prompts for improving AI-based text-to-image generation has remained a substantial yet unaddressed challenge. Hence we introduce GreenStableYolo, which improves the parameters and prompts for Stable Diffusion to both reduce GPU inference time and increase image generation quality using NSGA-II and Yolo. Our experiments show that despite a relatively slight trade-off (18%) in image quality compared to StableYolo (which only considers image quality), GreenStableYolo achieves a substantial reduction in inference time (266% less) and a 526% higher hypervolume, thereby advancing the state-of-the-art for text-to-image generation.