Improving Accuracy by Using Tailored Diffusion Model
Pengjun Zeng, D. Li, Shuangshuang Li · 2024
The images generated based on diffusion models have attracted global attention for their impressive ability to create convincing images. However, their complex internal principles and operations often make them difficult for non-professionals to understand and use. For a non-professional, pre-trained diffusion models are more user friendly. We introduce a diffusion model and trained model with different datasets, we analyze the performance of this model and find a strong relationship between prompts, dataset, and the accuracy of generated pictures.