DARNet: Deep Active Ray Network for Building Segmentation

This repository contains code for the DARNet framework as described in our CVPR 2019 paper.

Dependencies

• Python 3 (we used v3.6.5 with Anaconda)
• PyTorch (we used v0.4.1)
• PIL
• scipy and associated packages
• tqdm

Instructions

1. Download datasets from here
2. Unzip datasets to your desired directory
3. Modify setup.ini to reflect these directories, and the directory where you intend to keep results
4. Complete setup by running ./setup.sh
5. Run experiments with runner.py (modify as necessary to run a subset, or to coordinate across different machines, etc.)

Please contact me at dominic@cs.toronto.edu for questions.