Extracting Slums from Satellite Imagery¶
- 🔬 Data Science
- 🥠 Deep Learning and pixel-based classification
Table of Contents¶
- Export Training Data
- Train the Model
- Deploy Model and Extract Slum Boundaries
Every geographical area has its own definition for slum, but slum is usually an area with substandard housing and inadequate services. Slum is an important issue in developing countries as people reside in hazardous living conditions in congested slums and it makes them vulnerable to natural, physical and social disasters. Owing to the sensitivity of the slums, authorities need to continuously track and monitor the growth of the slums.
This sample shows how we can extract the slum boundaries from satellite imagery using the learn module in ArcGIS API for Python. The model trained here can be deployed on ArcGIS Pro as well as ArcGIS Enterprise and even support distributed processing for quick results.
In this workflow we will basically have three steps.
- Export Training Data using ArcGIS Pro
- Train a Model using learn module in ArcGIS API for Python
- Deploying the Model on ArcGIS Pro
For this sample we will be using data from Dharavi area in Mumbai, India. For a comparative analysis we will be using imagery for 2004 and 2014 acquired using quickbird and worldview3 sensors with a spatial resolution of 60cm.