There are a couple of goals:
- Create pipeline for hardcoded StableDiffusion prompt using name of food + list of ingredients, then pass into inversecooking model
- Directly use guided image-to-image StableDiffusion to do style transfer, run inference using inversecooking model
- Add in option to add other ingredients into StableDiffusion prompt
- Add in option to add both closest and farthest ingredient in latent space.
- Understand and visualize latent space that is learned from ingredient decoder.
utils.py: Contains all of the utility functions. Preprocessing, file saving/model loading, and anything regarding data format.
main.py: House CLI way of conducting model inference.
prompt.py: Defines class that generates the prompt to pass into StableDiffusion. Should be generalizable for range of prompts.
eda.ipynb: All of the exploratory data analysis.
utils.py
- A
main.py
- A
prompt.py
- A
eda.ipynb
- A