To replicate:

1. Start virtual environment
2. From project root directory run pip install -r /path/to/requirements.txt
3. Copy and paste input .tif file into project root directory
4. Call FieldDetector code from a separate script (ex: separate_script.py) by:
   • importing the class from the module from field_detector import FieldDetector
   • instantiating the class instance to a variable field_detector = FieldDetector()
   • calling the run method field_detector.run()
5. From project root directory run python3 separate_script.py

Notes:

I built a rudimentary agricultural field detector that takes 8-band geotiff images in UTM projection, calculates NDVI values, and outputs a geotiff in lat/long (EPSG 4326) projection.

• NDVI values will be high where there are plants and lower where they are absent.
• Pixels in the output images have a value of 255 where fields are located and 0 elsewhere.
• Includes client input to specify the threshold for determining which NDVI values should be considered as part of a field.

References:

https://en.wikipedia.org/wiki/Normalized_difference_vegetation_index

• NDVI = (NIR - R) / (NIR + R)
• NIR = Near Infrared
• R = Red
• Band order in given image
  • Band 1: Coastal Blue
  • Band 2: Blue
  • Band 3: Green I
  • Band 4: Green
  • Band 5: Yellow
  • Band 6: Red
  • Band 7: Red Edge
  • Band 8: Near Infrared (NIR)