urban-sciences-model

Requirements

• python3.8
• CUDA10.1
• torch 1.7.1
• torchvision 0.8.2
• segmentation_models_pytorch

How it works

pip install requirements.txt

git submodule update --init --recursive

Start notebook

jupyter lab # or jupyter notebook

Prediction

You can get inferenced PIL image below!

from inference import get_masked_pil_img
output = get_masked_pil_img("path/to/file.jpg")