ENVI Deep Learning at Work
The ENVI Deep Learning module is offered as an extension to ENVI for desktop applications and is built on the ENVI Task framework. This means that classifiers can be built once and run in any environment, whether that’s your desktop computer, on-premises servers, or in the cloud. To demonstrate how you can use this technology, here are a few real-world examples of customer problems that have been solved using the module.
Climbing wind turbines to inspect for damage is expensive and can be dangerous. The ENVI Deep Learning module was used in combination with UAV data to discern damage to turbine blades from lightning strikes that require repair verses dirt, paint chips, and damage from bird strikes that do not need to be repaired. The result has been improved safety and reduced costs and inspection times.
Extracting curved rows in agricultural fields is a difficult problem for traditional remote sensing algorithms. Since the ENVI Deep Learning module works like your brain to identify and extract features, a customer was able to easily determine plant rows. This information was then used to count crops, identify missing plants, and generate management zones for yield improvements.
When disasters strike, response time is very important. The ENVI Deep Learning module has been tuned so that you don’t need thousands of samples to create models for finding features. After a recent earthquake, the ENVI Deep Learning module was used to quickly identify rubble piles across a city. In this example, only 10 polygons were used to build an accurate classifier providing actionable information, so responders could navigate and move supplies in order to save lives.