Awesome

RoadTracer-M

This repository is the official implementation of the paper'ROAD NETWORK EXTRACTION FROM SATELLITE IMAGES USING CNN BASED SEGMENTATION AND TRACING'(IGARSS 2019).

Pipeline

Results

Requirements

To install requirements:

pip install -r requirements.txt

tensorflow
rtree
numpy

For generating the dataset, please refer to RoadTracer for more details.

Structure

• Corner/corner_detection.py : generating the starting points for further road tracing.
• Tracer/: tracing the road centerline from these starting points. Some of them are modified from RoadTracer.

Training

To train the model(s) in the paper, run this command:

python train.py --input-data <path_to_data> --alpha 10 --beta 20

📋 Describe how to train the models, with example commands on how to train the models in your paper, including the full training procedure and appropriate hyperparameters.

Contributing

• yao WEI
• kai ZHANG

Citation

If you find our work useful in your research, please cite:

  @INPROCEEDINGS{8898565,
  author={Wei, Yao and Zhang, Kai and Ji, Shunping},
  title={Road Network Extraction from Satellite Images Using CNN Based Segmentation and Tracing}, 
  booktitle={IGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium}, 
  year={2019},
  pages={3923-3926}
  }