The major threat that occurs in pipelines is leakage. The effects of leakage go beyond repair expense and cost of lost oil or gas. It also significantly affects the human lives and environment. In this white paper we have devised a Deep Learning-based solution using Convolution Neural Networks(CNN) and image processing techniques to detect and localize the leakages in oil & water pipelines.
What is in it for you?
The model successfully detects leakages in pipelines even in the presence of limited data. A Deep Learning model in general performs better when the amount of data available is large. Increasing the number of images would result in selecting the optimal hyper parameters for better model stability and reliability.