Home

Awesome

Multi-resolution-deep-neural-network

MR-DNN: is based on deep learning used to extract rice field from Landsat 8 satellite imagery. Using publicly available satellite imagery data we train a convolutional neural net to predict rice fields in satellite images. MR-DNN achieved tremendous performance for the prediction of rice from Landsat 8 satellite imagery data.

<img src= "https://user-images.githubusercontent.com/32522237/128299362-2e3a4403-793e-4a52-a503-5f59a40c1ee7.JPG" width= "256"> <img src= "https://user-images.githubusercontent.com/32522237/128299366-597380aa-6e88-493b-90c6-ed9584818c10.jpg" width = "256">

Data Downloading

You can download the satellite imagery from the (https://scihub.copernicus.eu/) or USGS Earth Explorer (https://earthexplorer.usgs.gov/). The following are the entity IDs of the images we used. To find images by their ID first select the right dataset (in our case Landsat 8 30m) and then go to "Additional criteria". Here are some IDs as an example we used:

LC08_L2SP_139044_20181001_20200830_02_T1

LC08_L2SP_148039_20190917_20200826_02_T1

Dependencies

  1. python==3.7.4
  2. tensorflow-gpu==1.13.1
  3. keras==2.2.4
  4. sklearn==0.21.2
  5. numpy==1.16.4
  6. matplotlib==3.1.1
  7. pandas==0.25.1