Text to image generation using stable diffusion / Muhammad Aizaq Azman

The results of a study on the text to image generation using stable diffusion are presented in this publication. The goal of the project was to create a system that could create a real human face based on the user description. The proposed system was developed in 3 phases consists of preliminary phase, design and implementation phase, and evaluation phase. The study utilized a dataset that has in LAION-5. The pre-trained model, v1-5-prunned-emaonly in hugging face is used as base model because this project applies transfer learning. The app is designed to be webpage by using the visual studio code. The model is evaluated by using 3 ratio train-test split performance and the highest performance and accuracy is chosen to be the final model.