logo

Projects

  • 2020
    • 𝗣𝗵𝗻𝗼𝗺 𝗣𝗲𝗻𝗵: 𝗨𝗿𝗯𝗮𝗻 𝗚𝗿𝗼𝘄𝘁𝗵 𝗳𝗿𝗼𝗺 𝟭𝟵𝟴𝟴 𝘁𝗼 𝟮𝟬𝟮𝟬 𝗯𝘆 𝗟𝗮𝗻𝗱𝘀𝗮𝘁 𝗦𝗮𝘁𝗲𝗹𝗹𝗶𝘁𝗲 𝗜𝗺𝗮𝗴𝗲𝗿𝗶𝗲𝘀
    • 𝗦𝗲𝗻𝘁𝗶𝗻𝗲𝗹-𝟭 𝗦𝗔𝗥: 𝗖𝗮𝗺𝗯𝗼𝗱𝗶𝗮 𝗙𝗹𝗼𝗼𝗱 𝗶𝗻 𝗢𝗰𝘁𝗼𝗯𝗲𝗿 𝟮𝟬𝟮𝟬
    • 𝗖𝗮𝗺𝗯𝗼𝗱𝗶𝗮 𝗙𝗼𝗿𝗲𝘀𝘁 𝗖𝗼𝘃𝗲𝗿 𝗖𝗵𝗮𝗻𝗴𝗲 𝟮𝟬𝟬𝟬-𝟮𝟬𝟭𝟵
  • 2021
    • 𝗗𝗲𝘃𝗲𝗹𝗼𝗽 𝗽𝗮𝗱𝗱𝘆 𝗮𝗿𝗲𝗮 𝗺𝗮𝗽 𝗳𝗿𝗼𝗺 𝗠𝗢𝗗𝗜𝗦 𝘀𝗮𝘁𝗲𝗹𝗹𝗶𝘁𝗲 𝗶𝗺𝗮𝗴𝗲𝘀 𝘂𝘀𝗶𝗻𝗴 𝗺𝗮𝗰𝗵𝗶𝗻𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴
      • 1. Export MODIS satellite images from Google Earth Engine
      • 2. Create Mosaic of 2011 NDVI Images
      • 3. Create DataFrame of NDVI Timeseries
      • 4. Noise Reduction in NDVI Timeseries
      • 5. Validation Data
      • 6. KMeans Clustering and Results
    • 𝗜𝗻𝗱𝗼𝗻𝗲𝘀𝗶𝗮 𝗽𝗿𝗼𝗷𝗲𝗰𝘁: 𝗵𝗶𝗴𝗵-𝗿𝗲𝘀𝗼𝗹𝘂𝘁𝗶𝗼𝗻 𝘀𝗮𝘁𝗲𝗹𝗹𝗶𝘁𝗲 𝗶𝗺𝗮𝗴𝗲 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀
      • 1. Extracting vegetation area
      • 2. Extracting water area
      • 3. Extracting building area
  • 2022
    • 𝗞𝗮𝗻𝗼 & 𝗬𝗼𝘀𝗵𝗶𝗶 𝗿𝗶𝘃𝗲𝗿: 𝗟𝗮𝗻𝗱 𝗰𝗼𝘃𝗲𝗿 𝗰𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝘂𝘀𝗶𝗻𝗴 𝗺𝗮𝗰𝗵𝗶𝗻𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴
      • 1. Prepare input data of Kano River
      • 2. Prepare input data of Yoshii River
      • 3. Supervised learning model development
      • 4. Predicting the whole Kano and Yoshii River
    • 𝗠𝗮𝗽𝗽𝗶𝗻𝗴 𝗳𝗹𝗼𝗼𝗱 𝗳𝗿𝗲𝗾𝘂𝗲𝗻𝗰𝘆 𝗶𝗻 𝗖𝗮𝗺𝗯𝗼𝗱𝗶𝗮 𝗳𝗿𝗼𝗺 𝟭𝟵𝟴𝟴 𝘁𝗼 𝟮𝟬𝟮𝟬 𝘂𝘀𝗶𝗻𝗴 𝗚𝗼𝗼𝗴𝗹𝗲 𝗘𝗮𝗿𝘁𝗵 𝗘𝗻𝗴𝗶𝗻𝗲
    • 𝗟𝗮𝗻𝗱 𝗰𝗼𝘃𝗲𝗿 𝗰𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝘂𝘀𝗶𝗻𝗴 𝘀𝘂𝗽𝗲𝗿𝘃𝗶𝘀𝗲𝗱 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗺𝗲𝘁𝗵𝗼𝗱 𝗶𝗻 𝗚𝗼𝗼𝗴𝗹𝗲 𝗘𝗮𝗿𝘁𝗵 𝗘𝗻𝗴𝗶𝗻𝗲
    • 𝗩𝗶𝘀𝘂𝗮𝗹𝗶𝘇𝗲 𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗳𝗼𝗼𝘁𝗽𝗿𝗶𝗻𝘁 𝗼𝗳 𝗖𝗮𝗺𝗯𝗼𝗱𝗶𝗮 𝗶𝗻 𝗚𝗼𝗼𝗴𝗹𝗲 𝗘𝗮𝗿𝘁𝗵 𝗘𝗻𝗴𝗶𝗻𝗲
    • 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 & 𝗡𝗲𝗮𝗿 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝗟𝗮𝗻𝗱 𝗨𝘀𝗲 𝗟𝗮𝗻𝗱 𝗖𝗼𝘃𝗲𝗿 𝗠𝗮𝗽𝗽𝗶𝗻𝗴 - 𝗖𝗮𝗺𝗯𝗼𝗱𝗶𝗮

Documentations

  • LiDAR
    • Create DEM and Hillshade from point cloud data in Python
    • Create RGB image from las file in Python
  • Geo-Python
    • Read & visualize raster image using xarray
    • Classify iris dataset with random forest classifier
    • Create a subplot figure
    • Interactive geoplots in dashboard layout with Bokeh
    • Convert GPS-tracked bus movement data from JSON to GeoJSON format
    • Create water occurrence image from yearly satellite images
    • Download forecast precipitation dataset from CHIRPS-GEFS website using Python
    • Calculate daily average potential evaporation from 3-year dataset for Source Model
    • Convert from Vector to Raster
  • Google Earth Engine
    • Download DEM from SRTM90 dataset
    • Calculate monthly mean temperature from ECMWF Climate dataset
    • Calculate monthly mean precipitation from CHIRPS Daily dataset
    • Calculate water turbidity index (WTI) from Sentinel-2
  • Machine Learning
    • Land Use Classification with Random Forests Classifier
  • Deep Learning
    • UNET Building footprint extraction using PyTorch
      • Convert WorldView-3 satellite images from Uint16 to 8byte
      • Convert building polygons to building masks for semantic segmentation model
      • UNet-Building Segmentation Model for WorldView-3 Satellite Images (PyTorch)
      • Split large satellite image into small images for model prediction
      • Building Footprint Prediction on WorldView-3 Satellite Images

Lessons

  • Geospatial data analysis with Python
    • Session 1: Geometric Objects
    • Session 2: Vector Data Analysis and Map Projection
    • Session 3: Geocoding and Nearest Neighbour Analysis
    • Session 4: Geometric operation and Data classification
    • Session 5: Plotting Static and Interactive Map on Leaftlet
    • Session 6: Raster Data Analysis
Theme by the Executable Book Project

UNET Building footprint extraction using PyTorch¶

Check out the codes belows:

  • Convert WorldView-3 satellite images from Uint16 to 8byte
    • Objective
    • Dataset
    • Code
  • Convert building polygons to building masks for semantic segmentation model
    • Objective
    • Dataset
    • Code
  • UNet-Building Segmentation Model for WorldView-3 Satellite Images (PyTorch)
    • Objective
    • Code
  • Split large satellite image into small images for model prediction
    • Objective
    • Code
  • Building Footprint Prediction on WorldView-3 Satellite Images
    • Objective
    • Code

previous

Deep Learning

next

Convert WorldView-3 satellite images from Uint16 to 8byte

By Men Vuthy
© Copyright 2022, Men Vuthy.