CSE Developer Blog

Classifying Leaks Using Cognitive Toolkit

May 10, 2017 May 10, 2017 05/10/17

Michael Lanzetta

We use Deep Learning to turn a painful and time-consuming leak-detection task for water and oil pipelines into a fast, painless process. Using Python and Fast Fourier Transforms, we turn audio sensor data into images, then use Convolutional Neural Networks to detect and classify pipeline anomalies.