Detection of electric utility features and vegetation encroachments from satellite images using deep learning¶

• 🔬 Data Science
• 🥠 Deep Learning and Object Detection

Table of Contents¶

• Introduction
• Necessary imports and get connected to your GIS
• Part 1 - Train the model and detect electric utility features
  • Export electric utility training data for deep learning
  • Load RetinaNet model architecture
  • Tuning for optimal learning rate
  • Fit the model on the data
  • Unfreeze and fine tuning (optional))
    • Unfreeze model
    • Optimal learning rate
    • Fit model on the data
  • Save the electric utility detection model
  • Load an intermediate model to train it further
  • Visualize results in validation set
• Part 2 - Train the model and detect trees
  • Export trees training data for deep learning
  • Load RetinaNet model
  • Finding optimal learning rate
  • Fit the model on the tree data
  • Save the tree detection model
  • Visualize results in validation tree dataset
• Part 3 - Deploy model and detect electric utility features & trees at scale
• Part 4 - Near analysis to find possible vegetation encroachment near electric utility features
• Conclusion
• References

Introduction¶

This sample notebook demonstrates how to efficiently map the electric utility features and trees in the imagery with possible locations of vegetation encroachment. Satellite imagery combined with machine learning leads to cost-effective management of the electric grids. This workflow consists of four major operations:

• Building and extracting training data for electric utility and trees using ArcGIS Pro
• Training a deep learning model i.e. RetinaNet using arcgis.learn
• Model inferencing at scale using ArcGIS Pro
• Proximity analysis between detected objects (electric utility and trees) feature layers using ArcGIS Pro