README.md

urban-sciences-model

Requirements

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

How it works

pip install -r requirements.txt

git submodule update --init --recursive

Preparation

You should set best_model_Unet_resnet50_epoch40.pth into notebooks directory, if you want to use a pre-trained model. Also you can get the weight from here.

Start notebook

jupyter lab # or jupyter notebook

Prediction

You can get the inferenced PIL image below!

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

Original paper

Quantifying urban streetscapes with deep learning: focus on aesthetic evaluation

Dataset

Citation

If this work is helpful for your research, please consider citing the following BibTeX entry.

@misc{kumakoshi2021quantifying,
      title={Quantifying urban streetscapes with deep learning: focus on aesthetic evaluation}, 
      author={Yusuke Kumakoshi and Shigeaki Onoda and Tetsuya Takahashi and Yuji Yoshimura},
      year={2021},
      eprint={2106.15361},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}