Skip to content

Urban change model designed to identify changes across 2 timestamps

Notifications You must be signed in to change notification settings

granularai/fabric

Repository files navigation

Code for the paper "Detecting Urban Changes with Recurrent Neural Networks from Multitemporal Sentinel-2 Data" (Accepted in IGARSS 2019)

This is the github repository containing the code for the paper "Detecting Urban Changes with Recurrent Neural Networks from Multitemporal Sentinel-2 Data" by Maria Papadomanolaki, Sagar Verma, Maria Vakalopoulou, Siddharth Gupta, Konstantinos Karantzalos.

MultiDate-LSTM code is available here.

Requirements

The code has been tested on:

  • 4xNvidia P100 GPU
  • Ubuntu 18.04 LTS on 96 vCPUs and 240 GB of RAM (Large scale inference will be slower on other configurations)
  • Pytorch v0.4.0
  • Opencv 3.0

Urbanization detection output

Multidate Sentinel-2 dataset

  1. OSCD
  2. OSCD + Our Dates
  3. Pretrained weight for Bi-Date Siamese Model

Contact

For any queries, please contact

Sagar Verma: [email protected]