Compute Land Use Land Cover (LULC) for any area of your choice, using the model trained in the lulc_classification_using_deep_learning repo.
This repo is built on the part 2 of the lulc tutorial in CCAI Summer school.
High level flow:
- Select region of interest and get the boundary of the region
- Get sentinel2 data (R,G,B) (preferably a median image for a time period of interest) for the region of interest
- Generate 64x64 tiles within the region of interest (as required by the model in this case)
- Run predictions (of the fine tuned Resnet50 model) on each of the tiles in the region.
- Visualize the predictions
I computed LULC of 2024 of Bangalore, India where I live:
TODO : I plan to compute change of LULC for a few Indian cities, to quantify the reduction in green cover in the cities, in the last few years.