Backend API
Experience the future of interior design with our revolutionary AI Interior Designer tool. Unleash your creativity as our generative models transform your ideas into stunning, personalized living spaces. Design a completly new space or replace the furniture with few-click action.
This project was developed during the course Interaction and Information Design at FRI.
API uses Stable Diffusion to generate stunning images. Stable Diffusion model is not run on your machine instead it is outsourced to the Novita.ai API. In order to run the backend service you need Novita.ai API KEY. Please paste the API KEY in the .env file. Sample env file (called .sample_env) is present in the repository. Copy this file and rename it to .env. To generate even better results Segment-Anything Model is added to segment any object in the area. You will need to download pre-trained SAM model weights (more on that in Installation section)
Since the backend is built using FastAPI you need python and pip installed on your machine. PyTorch and Torchvision packages are included in requirements.txt file.
- Clone the repository
- Move to the project's root folder
- (OPTIONAL) Create virtual environment
python3 -m venv venv - (OPTIONAL) Activate virtual environment
source venv/bin/activate - Install required packages
pip install -r requirements.txt
You will also need to download SAM model weights here and save it in the project's root folder.
To start the Backend API you simply run: uvicorn main:app --reload. The server is listening on port 8000 by default. The API documentation is located on /docs.