/DeepGlobe-Road-Extraction-Challenge

D-LinkNet: LinkNet with Pretrained Encoder and Dilated Convolution for High Resolution Satellite Imagery Road Extraction

Primary LanguagePythonMIT LicenseMIT

DeepGlobe-Road-Extraction-Challenge

Code for the 1st place solution in DeepGlobe Road Extraction Challenge.

Requirements

  • Cuda 8.0
  • Python 2.7
  • Pytorch 0.2.0
  • cv2

Usage

Data

Place 'train', 'valid' and 'test' data folders in the 'dataset' folder.

Data is from DeepGlobe Road Extraction Challenge. You should sign in first to get the data.

Train

  • Run python train.py to train the default D-LinkNet34.

Predict

  • Run python test.py to predict on the default D-LinkNet34.

Download trained D-LinkNet34