Serverless Remote Sensing analytics for an electric utility customer

This article discusses a serverless remote sensing analytics solution developed by AWS Professional Services for an electric utility customer to predict line-of-sight (LOS) of power lines and assets for post-Public Safety Power Shutoff (PSPS) event inspections.

Specifically, the article covers:

• Overview of power and utility companies' need for vegetation management and asset inspection
• Goals and use case to predict LOS of power assets from aerial imagery to support post-PSPS inspections
• Data preprocessing and modeling strategy integrating LiDAR, satellite imagery, and GIS data
• Model training and performance using ResNet-100 architecture on labeled imagery
• Deployment of serverless, automated, and scalable batch inference solution using AWS services like Lambda, SageMaker, and S3
• Visualization of inference results with GIS tools to identify LOS and NO-LOS line segments
• Conclusion highlighting the automated, scalable, and cost-effective solution for on-demand LOS prediction