Synthetic images can expedite machine learning building and training process. Obtaining real images that contain specific features is hard. In this talk we see how to generate synthetic satellite images by fine-tuning stable diffusion.
We have all seen how Machine Learning can be effectively used to build systems that can identify objects in a satellite image. However, obtaining images for trainings these systems is hard. In this talk, I will show you how to fine-tune stable diffusion to generate synthetic satellite images. I will show you a demo of generating satellite images containing highways by fine-tuning stable diffusion using dreambooth.
