The client had an existing system where Special Meter Readers would do monthly cycles of collecting readings from customer utility meters, the procedure would involve taking a picture of the meter along with a manual reading being recorded. The Meter Reader would then submit both the reading and the image to the server. A unit of Quality Assurance individuals would then manually compare the recorded readings against the readings shown in the image, in case of any discrepancy the Meter Reader would be notified. This QA process was very tedious, expensive and time consuming. To come up with an optimized solution, ByteCorp conducted an R&D cycle to create Deep Learning model that would record the readings from the collected image at runtime and compare with the manually entered readings, in case of a match the QA team would not have to manually screen it, saving time and money.