A Japanese based Instrumentation company wanted to develop and implement a Gas Leak Detection System using Image processing and Neural networks. They wanted to replace a high cost, hardware based solution with a scalable software and inexpensive hardware.
A RasberryPi along with thermal image camera, supported by a machine learning algorithms running on the Raspberry PI and a backend server was implemented to detect the leak and type of gas that is leaked. Solution was tested and certified by a government agency [FCRI] of India.